LibreOffice Calc Set To Get GPU Powered Boost From AMD

darthcamaro writes "We all know that the open source LibreOffice Calc has been slow — forever and a day. That's soon going to change thanks to a major investment made by AMD into the Document Foundation. AMD is helping LibreOffice developers to re-factor Calc to be more performance and to be able to leverage the full power of GPUs and APUs. From the article: '"The reality has been that Calc has not been the fastest spreadsheet in the world," Suse Engineer Michael Meeks admitted. "Quite a large chunk of this refactoring is long overdue, so it's great to have the resources to do the work so that Calc will be a compelling spreadsheet in its own right."'" Math operations will be accelerated using OpenCL, unit tests are being added for the first time, and the supposedly awful object oriented code is being rewritten with a "modern performance oriented approach."

Comments

If you need it you are doing it wrong.

[140Mandak262Jamuna]

If your spreadsheet needs a gpu to speed up calculations, you are probably misusing spreadsheets. I know most accountants love the spreadsheet and they make insanely complicated things using spreadsheets pushing it far beyond what these are designed to do. But if you have a spreadsheet that needs this much of cpu time to recompute, you should probably be using a full fledged data base with multiple precomputed indexing.

Re:If you need it you are doing it wrong.

[Russ1642]

Custom database applications are expensive and inflexible. Stop trying to tell people what they can't do with a spreadsheet.

Re:If you need it you are doing it wrong.

[buchner.johannes]

I agree. Also, if you rewrite structured code into a "performance oriented approach", you are doing it wrong.
Write code so it is easy to understand. Then compilers should understand how to make it fast.
This can only come from people who think code is for machines. Code is for humans to read and modify.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

Spreadsheets are all rectangular. That's pretty inflexible. Show me a triangular spreadsheet and then we'll talk.

Re:If you need it you are doing it wrong.

[Lunix Nutcase]

Then compilers should understand how to make it fast.

Should but often don't.

Re:If you need it you are doing it wrong.

[Impy the Impiuos Imp]

Object-oriented code, not structured code.

For all I know every cell is an instantiated object held in a giant linked list, with every property another object.

It may be a fantastical design, but with real-world efficiency problems.

Re:If you need it you are doing it wrong.

[Kenja]

Pivot Tables can have three or more axis.

Re:If you need it you are doing it wrong.

[BitZtream]

Thats not the issue. If your spreadsheet is SO larger that on a MODERN CPU, its slow ... you're doing it wrong.

You can make insanely complex, application like spreadsheets, without noticing 'recalc' time. By the time you get to noticing 'recalc' time, you've fucked up.

Caveat: OO.org is known to have some of the crappiest code in existence, so with the case of Calc, you don't have to make ridiculous spreadsheets to notice recalc time. GPU support won't fix the problem however as its not the math thats the issue, its the shitty logic code filled with stupid crap written by clueless devs that cause Calc to be so slow.

Re:If you need it you are doing it wrong.

[symbolset]

So you haven't heard of Microsoft Excel for supercomputers then? What about the Excel RPG?

Re:If you need it you are doing it wrong.

[geminidomino]

GPU support won't fix the problem however as its not the math thats the issue, its the shitty logic code filled with stupid crap written by clueless devs that cause Calc to be so slow.

Indeed. You really shouldn't need to have to get a gaming GPU to run a spreadsheet. Hopefully

"and the supposedly awful object oriented code is being rewritten with a "modern performance oriented approach".

means they intend to address that part, too, and the crappy headline is just being whiz-bang. If they're dumb enough that they think throwing inappropriate hardware at the problem is a solution... well, they're too far from the vicinity of the US Pacific coast...

Re:If you need it you are doing it wrong.

[Anonymous Coward]

I take it you have not tried to write an optimizing compiler any time recently, then?

HLLs makes code easier to read. HLLs also makes it easy to generate non-optimal code. Writing your HLL _just so_ might improve performance tremendously. Example: which way do you iterate through an array to apply an operation to each element? Well, _if_ you happen to have iterated through it recently, doing it backwards might be lot faster then doing it forwards due to cache locality. I'd love to see a compiler managing to take advantage of that.

Compilers are lovely, but they can only take advantage of fairly simple things. You need to write your code and data structures so that the result isn't a twisted mess that runs like three-legged dog on sedatives.

Re:If you need it you are doing it wrong.

[wonkey_monkey]

If you need it you are doing it wrong

That's begging the question, sort of. Who said anything about needing it?

If your spreadsheet needs a gpu to speed up calculations, you are possibly misusing spreadsheets.

FTFY. If a spreadsheet is capable of doing what someone wants, who are you to say it shouldn't be done that way?

But if you have a spreadsheet that needs this much of cpu time to recompute

Again with the "need." This isn't being done for the people who need fast spreadsheets. It's being done so all spreadsheets can go faster. Who wouldn't appreciate a spreadsheet recalculating in 0.1s instead of 0.5s?

you should probably be using a full fledged data base with multiple precomputed indexing.

Well, now you can draw your arbitrary "this is too slow for spreadsheets" line further away from Calc. That's all.

Re:If you need it you are doing it wrong.

[camperdave]

This! I once wrote a calculator program for a microprocessor programming course that employed the fact that multiplication is just repeated addition, and addition is just repeatedly incrementing one number while decrementing the other. It worked, but boy, was it dog slow.

Re:If you need it you are doing it wrong.

[fuzzyfuzzyfungus]

An N-dimensional spreadsheet probably wouldn't be too hairy to describe as a mathematical structure; but the UI might get pretty dreadful.

Re:If you need it you are doing it wrong.

[wonkey_monkey]

Write code so it is easy to understand. Then compilers should understand how to make it fast.

Could a compiler have come up with the fast inverse square root?

I once got a 10% speed increase just by moving a pointer offset increment. The compiler missed that one.

Re:If you need it you are doing it wrong.

[gstoddart]

you should probably be using a full fledged data base with multiple precomputed indexing

Well, you can put together a spreadsheet in a few hours.

What you're describing is likely months of custom development and design, and a whole new thing to maintain.

Spreadsheets are popular because they're easily deployed, don't require any extra licensing, and the people who know how to use them can likely do things with them that some of us would be astounded at.

I know people who use spreadsheets for pretty much everything, because it's available to them readily, and they've been using them for a long time.

It's all well and good to suggest that they use a full-fledged database -- but in reality, they can probably get something useful in a few days for a fraction of the cost.

It sounds like in this instance, the code was just horribly inefficient.

Re:If you need it you are doing it wrong.

[Guy Harris]

I agree. Also, if you rewrite structured code into a "performance oriented approach", you are doing it wrong. Write code so it is easy to understand. Then compilers should understand how to make it fast.

I.e., code should be written in high-level descriptive languages, and the compiler should choose the algorithm, so that a tricky-to-understand but much-faster algorithm doesn't show up in the code as written, but shows up in the generated code?

Not all rewrites-for-performance involve low-level trickiness.

Re:If you need it you are doing it wrong.

[fermion]

In the late 80's I developed some relatively complicated models on the first version of Excel. The big concession I had to make was turning autocalc off. The longest process was actually printing the report. And, btw, this including exporting and importing data into a database outside Excel. This was on an original Mac. A few years later I was working with huge telemetry data sets that had to be scrubbed, reorganized, and plotted. I had been on the spreadsheet binge for years, so I started off with some C++ for scrubbing then importing it into Excel for organization and graphing. After a couple days I thought how stupid this is, found gnuplot, and had it done, so, once again the longest task was waiting for the pen plotter to finish the pretty pictures. So yes, it is a matter of the right tool for the job. Of course so many people only know how to use a spreadsheet, so asking them to use a more efficient tool is out of the question. In reality clock cycles are infinitely cheaper than people, so it is often good to figure our how to maximize the clock cycles rather than change the people. Look at the GUI interface. An incredible waste of clock cycles. But for many things it allows the use of cheaper less qualified people, and sometimes allows more expensive people to work more efficiently. So win-win.

Re:If you need it you are doing it wrong.

[robthebloke]

I agree. Also, if you rewrite structured code into a "performance oriented approach", you are doing it wrong.

Nonsense. One of the joys of C++, is the lack of reflection. This tends to lead apps down the route of wrapping everything into an 'Attribute' class of some description, and wiring those attributes together using a dependency graph. The problem with this (very clean OOP) approach, is that it simply doesn't scale. Before too long, this constant plucking of individual data values from your graph, ends up becomming a really grim bottleneck. If you then run the code through a profiler, rather than seeing any noticeable spikes, you end up looking at an app that's warm all over. If you're in this situation, no amount of refactoring is going to save the product. You're only option is to restructure the

The "performance oriented approach" is the only approach you can take these days. Instead of having a fine OOP granularity on all of your data, you batch data into arrays, and then dispatch the computation on the SIMD units of the CPU, or on the GPU.

Then compilers should understand how to make it fast.

Uhm, nope. Sure, if you happen to have 4 additions right next to each other, the compiler might replace that with an ADDPS. In the case in point however, you'll probably expect a generic node to perform the addition on single items in the table. As such, your "addTwoTableElementsTogether" node isn't going to have 4 floating point ops next to each other, it will only have one. Compilers cannot optimise your data structures. If you want to stand a chance of having the compiler do most of the gruntwork for you, you actually have to spend time re-factoring your data structures to better align them with the SIMD/AVX data types. Some people call this a "performance oriented approach".

This can only come from people who think code is for machines. Code is for humans to read and modify.

Bullshit. This can only come from experienced software developers who understand that the only approach to improving performance of a large scale app, is to restructure the data layout to better align it with modern CPUs. There is *NOTHING* about this approach that makes the code harder to read or follow - that's just your lack of software engineering experience clouding your judgement.

Re:If you need it you are doing it wrong.

[Chatterton]

Done the same during the algorithm course. On a turing machine you don't have such shiny operators like addition even less multiplication :)

Re:If you need it you are doing it wrong.

[gstoddart]

I agree. Also, if you rewrite structured code into a "performance oriented approach", you are doing it wrong.
Write code so it is easy to understand. Then compilers should understand how to make it fast.

Except code can end up going through so many layers of abstraction, with some of those layers doing things in the most inefficient manner possible because terrible assumptions were made.

Sometimes, you need to plan for both performance and well structured code -- or you can end up writing garbage which makes heavy use of code which does extremely stupid things.

Most people nowadays are so far removed from knowing what's happening close to the metal that they don't often realize you're essentially running thousands of lines of code to do something trivial as heavyweight libraries wrap everything up.

At a certain point, someone needs to be able to determine if the code is the reason your stuff is so damned slow, and if the bloat has gotten out of hand. Sometimes, you really do need to look at your own optimizing.

Re:If you need it you are doing it wrong.

[Charliemopps]

You are correct for business applications. But often what spreadsheets are used for is: "I need this quarterly report figured out for the meeting on Friday and then I'm going to delete it forever." Going out an building a full fledged database for that would be stupid. Having a very complicated spreadsheet that solves a problem isn't bad... using that spreadsheet over and over as part of your business process is.

Re:If you need it you are doing it wrong.

[jellomizer]

Spreadsheets are good for "throwaway applications" you need to do these calculations fast or gather data, and after a few weeks you don't need it anymore.
If you are going to be following a process with a fairly rigid data sets. You are going to be better off spending the time and money to make a real application with a real database with it. That way the rigidness is to your favor to prevent incompatible creep, and allow for future data gathering abilities.

Using Spreadsheets for your application needs works but it is very flimsy and over the long run you will be spending a lot more time fixing your mistakes (say a bad sort) Or a mistime change and save, or just the wrong click of your mouse you messed up a lot of data.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

How is a triangle not two-dimensional?

Re:If you need it you are doing it wrong.

[tibit]

Probably they had some developer mobility betwen Star Division and SAP :/ /me ducks and runs

Re:If you need it you are doing it wrong.

[tibit]

If so, it probably needs a custom allocator, that's all.

Re:If you need it you are doing it wrong.

[squiggleslash]

If the tool you're using isn't an exact fit for the application you need it to do - though only because of speed, but it does it anyway, you should use another tool even though...

1. You're not familiar with it
2. It's not your area of expertise
3. The lack of suitability means you're wasting whole minutes a day. However, you'll need to hire a programmer or else take a course likely lasting months to get the skills needed to use the alternative tool.
4. Your tool, the one that was designed to be easy for someone in your profession to make use of, does, ultimately, do the job you need it to do.

No.

No, no, no, no and again no. You're a developer. Good for you. Good for me too. But our jobs are not to make patronising unrealistic suggestions to smart people who don't have our particular skillset. Our job is to make it easier for other people to do their jobs. Telling them to hire programmers or run off and learn our skills isn't "making it easier".

If the only thing wrong with the way they're doing is that their computer runs slow when doing it, the solutions we should be presenting are the easiest. Upgrades and better underlying office suite code. Good for the LibreOffice people that that is what they're doing.

Re:If you need it you are doing it wrong.

[tibit]

I wholly agree.

Re:If you need it you are doing it wrong.

[CastrTroy]

No, because then it would have been giving the wrong answer most of the time. The fact that the wrong answer is good enough in certain situations is not something the compiler would be able to determine.

Re:If you need it you are doing it wrong.

[tibit]

Months to put together a bit of SQL and some front end for it, using, say, oh horror of horrors, Excel? Next you're going to tell me that it takes a man-month to write a hello world.

Re:If you need it you are doing it wrong.

[gstoddart]

Next you're going to tell me that it takes a man-month to write a hello world.

Not at all, but I will flat out tell you that I've seen domain specific spreadsheets which have surprisingly little to do with adding numbers, and which if you tried to replace it with a DB application would take you months (or years) to do -- and you'd end up with something you still have to maintain.

Spreadsheets have the really nice feature of still mostly working when you upgrade the version of the software.

I'm not saying they're always the best solution, but surprisingly often it covers the "good enough and within budget" requirements.

I suspect in a lot of corporations if you handed down an edict saying "thou shalt not use spreadsheets for the following things", stuff would grind to a halt for a little while, and everyone would say the hell with it and go back to using them because they work well enough and are readily available.

Don't underestimate just how much business critical stuff is managed in spreadsheets. You might be shocked.

Re:If you need it you are doing it wrong.

[Chelloveck]

Example: which way do you iterate through an array to apply an operation to each element? Well, _if_ you happen to have iterated through it recently, doing it backwards might be lot faster then doing it forwards due to cache locality. I'd love to see a compiler managing to take advantage of that.

I'd trust the compiler to do it a whole lot more than I'd trust the programmer. Things like cache locality are going to be very dependent on machine architecture. You should not be optimizing for machine architecture on anything that's intended to be the least bit portable. Yeah, you made it faster on your machine, but what about your user's? Will it have the same cache size? The same line size? Hell, will it even have the same word size and endedness? To the extent that any of this should be optimized, it's the compiler's job to do it.

Be sure to pick your algorithms carefully. Know the complexity, and don't choose an O(n^2) algorithm when there's an equivalent one that's O(n). But optimizing hardware-level stuff? Unless you get to pick the hardware, don't do it. Every optimization you make on your machine is likely to be a pessimization on someone else's. The compiler knows the architecture it's compiling for (well, usually...) and is in a much better position to make such performance decisions.

Re:If you need it you are doing it wrong.

[gstoddart]

You're a developer. Good for you. Good for me too. But our jobs are not to make patronising unrealistic suggestions to smart people who don't have our particular skillset. Our job is to make it easier for other people to do their jobs. Telling them to hire programmers or run off and learn our skills isn't "making it easier".

This. A thousand times this.

Somewhere along the way, our industry has developed a collective mentality "we're smarter than you, and we will give you what we want even if we have no idea of what you need".

Once you get a little further removed and realize that the stuff we're writing/supporting is intended to help the people who do the real, bread and butter parts of the business -- you start to realize if we're an impediment to them, it's worse than if we weren't there at all.

They're not interested in some smug little bastard looking down his nose at them because they couldn't possibly do what he does. They're interested in getting their stuff done as quickly as possible.

I can tell you there is nothing more frustrating and counterproductive than some kid straight out of school who thinks the world needs to bow at his feet and stand aside to allow him to tell them how they should do things. Sadly, I've also met developers who have been in the industry a long time who still act like that.

In many industries, the people who do the real work of the company have highly specialized knowledge, and software is just a tool. And that tool is either helping them get stuff done, or frustrating the hell out of them.

Acting like we know better than they do (when we in fact know nothing at all about their domain expertise) is at best condescending, and at worst an impediment and a liability.

Re:If you need it you are doing it wrong.

[Jawnn]

It's all well and good to suggest that they use a full-fledged database -- but in reality, they can probably get something useful in a few days for a fraction of the cost.

It sounds like in this instance, the code was just horribly inefficient.

And if that were the end of it, we'd be in agreement. Delivering effective tools to users is what made the PC take off. That has not changed. Nevertheless, there are those who will, usually unwittingly, misuse those tools and, predictably, want to know why it doesn't work anymore. The fact is that there most certainly are times when a more sophisticated approach is called for.

Re:If you need it you are doing it wrong.

[fph il quozientatore]

Sure, it's all easy and fun until something like this happens: http://theconversation.com/the-reinhart-rogoff-error-or-how-not-to-excel-at-economics-13646. (I'm not saying that errors do not happen with databases, but the fact that the logic in your code is written in one bazillion copy-and-pasted formulas makes it very, very easy to screw up something. And it makes it impossible to write proper tests.)

Re:If you need it you are doing it wrong.

[aztracker1]

I've seen similar issues in higher level language based projects as well. For example in a simulation engine, each avatar in the simulation was given it's own thread... Now you add a few dozen avatars in a simulation, and more than a few dozen simulations running on a server, and you see the issue. The solution, have the engine run one thread that calls each avatar's "do-work" method in a loop. Minimal code change, huge performance gains for the server as a whole. Why? because context switching from one thread to another has a cost on the CPU. Not to mention a lot of lock(obj){} states that weren't needed with the loop.

It really bugs me when people think OO is the only way to get stuff done. Now, some functional language stuff just makes my head hurt, but there are plenty of functional patterns that can be used that will perform a lot better than OO, and still be just as readable.

Re:If you need it you are doing it wrong.

[lister king of smeg]

i thought excel for super computers was a joke...

Re:If you need it you are doing it wrong.

[gstoddart]

The fact is that there most certainly are times when a more sophisticated approach is called for.

Absolutely. But how often are companies willing to pay for them? In my experience, probably far less than you might feel they should.

You say 'musing the tools', they say 'getting their job done with the tools provided to them'.

There's all kinds of things which could or should be written as a specific application -- but getting most companies to invest in them is an uphill battle.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

Yes, we should all simply be content with what we have now and stop all progress.

Re:If you need it you are doing it wrong.

[unixisc]

Except that most lay people would find pivot tables and similar techniques easier to do on spreadsheets than learn database query languages, or whatever it would be that they need to get the job done.

Re:If you need it you are doing it wrong.

[lister king of smeg]

yes but we are talking about having to gpu accelerate a f'ing spreadsheet here, one a sheet gets to this point you are far and beyond the point where it is reasonable to switch to a more powerful tool. spreadsheets like this are huge amounts of spaghetti code mixed with badly written vb8 macros this is a nightmare for everyone involved. while this does not call for a full fledged compiled app, usually, switching over to using something like MS access or libra/open office base even if it means a few afternoons watching youtube walkthroughs would be a much better plan.

as for others saying that spreadsheets are compatible between versions. no because the macros used in these abominations are not always compatible.

Re:If you need it you are doing it wrong.

[TemporalBeing]

If your spreadsheet needs a gpu to speed up calculations, you are probably misusing spreadsheets. I know most accountants love the spreadsheet and they make insanely complicated things using spreadsheets pushing it far beyond what these are designed to do. But if you have a spreadsheet that needs this much of cpu time to recompute, you should probably be using a full fledged data base with multiple precomputed indexing.

The problem is that unlike MS Office, the ODF format does not typically store pre-calculated values of formals; so when you load a file it has to run all the formulas, etc to generate what the user wants to see. There's nothing wrong with that, it's just a little slower. An easy optimization would be for ODF to store formula results as part of the cell contents - but that's a change to the ODF standard AND the software that uses it.

Now, as per auto-calculating the spreadsheet and storing the values - that has its own issues, namely if the version of software that stored it had a bug something doing the calculation (much like Excel does is a good chunk of their mathematical functions, e.g. floor() is not the Mathematical floor() function you're use to, but Microsoft's variant of it) then that will get carried over and something will likely blame the wrong software for the bug when that software recalculates it the right way.

Note: The issues with Excel in that respect were well documented as part of the OOXML debacle - floor() doesn't work correctly on negative number IIRC.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

When it has 3 90deg angles being on the surface of a sphere.

Re:If you need it you are doing it wrong.

[nickittynickname]

That fair if your company doesn't have in house developers or your developers are completely saturated in projects. Otherwise you are most likely wasting time and money on something that can be developed quicker, be better maintained, and less error prone through other solutions.

Re:If you need it you are doing it wrong.

[orlanz]

For many of the situations that the parent is talking about, this is not true. Spreadsheets with massive business logic are extremely expensive and very inflexible, more so than DB apps. Just no central group/organization reviews, audits, and tallies these costs like they do for developed applications. Therefore, people assume the spreadsheet is cheaper. Do actual IT audits where these things fall into scope... and you quickly realize just how ridiculously risky the entire deck-of-cards-business is running.

Re:If you need it you are doing it wrong.

[dinfinity]

But if you have a spreadsheet that needs this much of cpu time to recompute, you should probably be using a full fledged data base with multiple precomputed indexing.

I agree, by that point, you should have definitely moved to MS Access.

Re:If you need it you are doing it wrong.

[tibit]

I know and it's lamentable. I still don't see what kind of an easy-to-do-in-a-spreadsheet kind of a business "application" would take months or years to do data storage/management in SQL...

Re:If you need it you are doing it wrong.

[equex]

Yeah we said this about gpu rendering of the UI as well....

Re:If you need it you are doing it wrong.

[garyebickford]

Lotus had a cool object-based spreadsheet system on the Next computer, called Improv. Improv attempted to redefine what spreadsheets were and how they worked, and once you got used to it, it was great. The basic principle was separation of data, views, and formulas (was this pre-figuring MVC?), and individual sheets could be any size - I'm not sure about 'triangular' per se though.

But alas, Improv never sold well on either NextStep (although it was very popular amongst financial modeling folks, and sold a lot of Nexts) or Windows 3.1 (W31 was probably way underpowered for the job). According to the Wiki article above, there are some descendants still in use.

Re:If you need it you are doing it wrong.

[gstoddart]

I still don't see what kind of an easy-to-do-in-a-spreadsheet kind of a business "application" would take months or years to do data storage/management in SQL...

Who said "easy to do"? I've seen stuff done in spreadsheets which evolved over years, and I know people who do stuff in Excel that leaves my head spinning.

You'd be amazed at the wacky stuff people can do in Excel. Many of the things I've seen done with it over the years probably evolved over a long time, but the end result is something which isn't trivial to replace without some pretty major investment.

It's just that you can start working in it and incrementally improve it to the point that it's almost a full on application in its own right.

Don't make the mistake of thinking everything people do in Excel is easy to do and you could quickly knock together a replacement. There's a lot of functionality in Excel, and those who have been using it a long time can do some pretty sophisticated things.

You make not like Excel, but don't go around thinking everything which can be (and is) done in Excel is trivial stuff. There's surprising amount of logic and conditional stuff you can build into it that most of us don't even know is there.

Re:If you need it you are doing it wrong.

[garyebickford]

The reality (sad ?) is that a huge percentage of financial modeling by the investment banks is done on Excel. Every trader has their own custom versions. Some flash crash-type events have been traced back to bugs in the spreadsheets. Excel makes it very easy to build a useful financial model very quickly, without a lot of 'programming' - although it is still programming, just don't tell those guys. (It's been pointed out elsewhere here that Lotus Improv and its descendants were/are better, and the descendants are still in use).

A book on functional programming pointed out that spreadsheets are a pretty good example of functional programming. I forget the details.

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

Spreadsheets are all rectangular.

That's not necessarily correct. In its core, a spreadsheet is a DAG of computation-performing nodes. The presentation style is merely a historical happenstance.

Re:If you need it you are doing it wrong.

[garyebickford]

I'm not a stock trader, but if I were, I'd much rather bang out a spreadsheet in 20 minutes (or modify one I'm already using) than wait several days at best for 'some dork in IT' to get a round tuit. Assuming it's possible to explain what I want to the IT guy*. In fact, IT will probably _never_ get a round tuit - they've got bigger fish to fry, and they're six months or two years behind schedule already. And the problem I'm working on will be over and done in 20 minutes - I need the model NOW, before London closes.

*I've got a friend in risk management at a big investment bank. I'm more up to speed on economics and finance than most geeks, and he twists my head around within the first two minutes, talking about differential velocity on secondary options, or something. It's basics to him but would require a week or two for me to figure out what he's talking about well enough so I could actually write the code he actually needs.

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

If they're dumb enough that they think throwing inappropriate hardware at the problem is a solution

What inappropriate hardware? OpenCL, for example, is quite flexible at allowing you to distribute problems to appropriate computing nodes in a heterogeneous system.

Re:If you need it you are doing it wrong.

[garyebickford]

You could have gone further - every arithmetic and logical operation can be performed by (IIRC) a series of XORs and NOTs. I look forward to your implementation, preferably using fluidics! :D

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

That's actually not a problem with the concept of a spreadsheet application, that's a problem with one particular spreadsheet application that has a horrible interface.

Re:If you need it you are doing it wrong.

[garyebickford]

Indeed. In my experience the only language that (historically) used more memory and cycles to get things done than OO was LISP - and I think that was largely corrected in later implementations of LISP. I have a project just now that uses a canned, very OO library to process (oddly enough) Excel spreadsheets. One particular sheet, with about 3000 mostly empty rows of data (no formulas), results in running out of memory after 20 minutes and 8 GB. The cause? I think each cell results in multiple separate objects being created.

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

Object-oriented code, not structured code.

These two are not complementary. Or did you throw away block-structured conditionals and loops in your OO code and reverted to using conditional and unconditional jumps?

Re:If you need it you are doing it wrong.

[Rockoon]

I still don't see what kind of an easy-to-do-in-a-spreadsheet kind of a business "application" would take months or years to do data storage/management in SQL...

The kind a mid-level employee evolves on a daily basis. Duh.

The task at hand isnt concrete, the task at hand is answering "What if I..."

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

HLLs makes code easier to read. HLLs also makes it easy to generate non-optimal code.

Except for many application areas where (V)HLLs can actually be more efficient than other languages. For example, some languages (such as Haskell) give you the nice property that nested maps can be fused.

Example: which way do you iterate through an array to apply an operation to each element? Well, _if_ you happen to have iterated through it recently, doing it backwards might be lot faster then doing it forwards due to cache locality.

This is 1) nonsense and 2) exactly the kind of code where a HLL will help you much more. (Why would you iterate twice through it and not use the loops in the way I've just described?)

Compilers are lovely, but they can only take advantage of fairly simple things.

Yeah, *simple* compilers, perhaps.

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

Could a compiler have come up with the fast inverse square root?

The answer is "probably yes". Granted, you'd have to design a compiler of your own - a very specialized one, one that would take numerical approximations and algebraic identities into consideration - but it's definitely not impossible, it's just that nobody probably felt it necessary to do so. There are also techniques out there for exploratory automatic programming (for example, genetic programming and superoptimization) that might help you with the problem.

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

The fact that the wrong answer is good enough in certain situations is not something the compiler would be able to determine.

You make it sound as if there are no accuracy metrics for numerical algorithms that can be automated. I don't think that this is the case.

Re:If you need it you are doing it wrong.

[K. S. Kyosuke]

This tends to lead apps down the route of wrapping everything into an 'Attribute' class of some description, and wiring those attributes together using a dependency graph. The problem with this (very clean OOP) approach, is that it simply doesn't scale. Before too long, this constant plucking of individual data values from your graph, ends up becomming a really grim bottleneck.

By this time, you are supposed to realized that you've essentially written a horribly complicated interpreter and that you ought to write a JIT compiler from the dependency graph into something more reasonable, for example OpenCL.

Re:If you need it you are doing it wrong.

[jeremyp]

It really bugs me when people conflate OO with anti-pattern x. In what world does OO mean the same as "use lots of threads"?

Re:If you need it you are doing it wrong.

[hedwards]

How is a compiler going to know which statement is most likely to cover a given case. Or that you really don't need a long to store that particular variable?

The point is that compilers optimizing is all well and good, but compilers aren't very good at identifying when you don't need a variable of a given size or in which order you should have your cases and if statements ordered for best performance.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

Actually, this doesn't work when implementing interpreters.
That's the reason, Mike Pall used assembly for LuaJIT.
http://article.gmane.org/gmane.comp.lang.lua.general/75426

Re:If you need it you are doing it wrong.

[Anonymous Coward]

I thought Excel for any computer was a joke.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

I have no problem with accountants using spreadsheets for complex things. They understand spreadsheets.

I do have problem with anyone associated with IT using a spreadsheet for a database.

Re:If you need it you are doing it wrong.

[jedidiah]

It's not 1998. Databases don't cost $60,000 anymore. There's no good reason that some cowboy developed database app should be any more expensive than some cowboy developed spreadsheet app.

At least the database allows for some semblance of centralized management.

Re:If you need it you are doing it wrong.

[SuiteSisterMary]

Sure, but I think the idea is that if your spreadsheet doesn't do all the calculations it needs to do more or less instantly, on any computer built within the last ten years, you're either a) horribly misusing your spreadsheet, or b) using a horribly written spreadsheet program.

It's kinda like redesigning your VW Bug to be able to back up to a full-on commercial loading dock to load groceries.

Now that said, sure, why the hell not? The GPU is just sitting there, might as well make use of it. But it's more along the lines of 'why the hell not' than 'this will make it faster!'

Re:If you need it you are doing it wrong.

[jedidiah]

That's fine for the first iteration of a prototype but not really maintainable or auditable after that.

One would hope that the total lack of any auditing would be considered a problem in a financial services environment.

Cowboys are for cattle country...

Re:If you need it you are doing it wrong.

[hairyfeet]

Oh Lord, preach brother preach! Do you know how many damned "Excel sheets from hell" I ran into when I was working corporate? That and "Access DB from the deep" were usually the 2 most abused tools I have EVER seen in all my days!

Now don't get me wrong, both spreadsheets and small DBs like Access can be great...if you know their limitations, hell I use an old VB into Access DB setup to catalog my DVDs, but what some of those guys were doing should have been a criminal offense. You'd have so much programming jammed into a spreadsheet it was unreal, or an entire office all depending on a single Access DB that had grown into this bloated monster.

So I have to agree, if your spreadsheet has become such a bloated mess that you actually need to offload to the GPU just to speed up all the data crunching that thing is doing? You might want to look into using something else instead of trying to get a spreadsheet to do it all.

I do have to say though that since taking over the guys at ODF seem to be doing a pretty damned good job modernizing LO, they keep this up and in another year and a half it'll probably be a drop in replacement for MS Office so great job guys.

Re:If you need it you are doing it wrong.

[Mr Z]

I'd like to see you implement an AND gate with just XORs and NOTs. Use as many as you like... ;-)

In all seriousness, you can build all other gates from NANDs. Or from NORs. In fact, that's how the Cray 1 was built, apparently. Tons and tons of ECL NAND gates.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

Nvidia CUDA acceleration add-ons do exist for Excel.

Re:If you need it you are doing it wrong.

[hairyfeet]

Dude VB into a DB (or whatever LO does into a DB, what language does BASE have?) is as old as the hills and most that I have seen has been a dozen times better than jamming everything into a spreadsheet.

Besides lets cut the bullshit and speak some truth, the whole reason spreadsheets got so abused and bloated was the bastard admins from hell wouldn't let anybody have an IDE or any tools EXCEPT for that copy of MS Office (or LO in this case) and as usual when you have a job to do and the admin is cockblocking you find a way around him and that was abusing the hell out of Excel and Access, or in this case Calc and BASE.

But I have found if the admin is reasonable and not an asshole and lets them have decent tools if they need them the spreadsheets from hell just doesn't happen, I would just ask guys what they were doing and if they had a legitimate plan to create something they required i was fine with letting them have an IDE like Netbeans or whatever they needed. I was always "the right tool for the job" kinda guy and I found if you just listened to the grunts and gave them the right tools when they needed them all those half assed spreadsheets and DB frontends just didn't happen, at least on my watch.

Re:If you need it you are doing it wrong.

[Zero__Kelvin]

"The fact that the wrong answer is good enough in certain situations is not something the compiler would be able to determine."

Sure it can; you simply have to tell it, by letting the compiler know that information just like you let it know everything else: ... wait for it ....in the source* ! .

* I am also referring to the build system makefiles, etc. since in 2013, if you check out the source code you get the makefiles, etc as part of the source

Re:If you need it you are doing it wrong.

[hairyfeet]

This is why I've been telling folks "Don't judge LO yet, they got a LOT of legacy cruft they are gonna have to remove so for a project THAT big give them around 2-3 years to get it all cleaned up" and I have to say they are making pretty good progress so far.

But you are right as far as CPUs go, I have an engineer doing huge Solidworks model presentations on a Phenom I triple and use a Bobcat dual core netbook as an HTPC running 1080P so if your spreadsheet manages to drag down even a 5 year old chip? You are doin it wrong.

Re:If you need it you are doing it wrong.

[garyebickford]

I knew it was something like that! :D It's been a couple of decades since I thought about this topic. I stand corrected.

Re:If you need it you are doing it wrong.

[garyebickford]

Yep, it's often a problem. :D And I'm only partially defending it, but it does have a place (and like any tool, should not be used in other places where it doesn't belong.) Also, IANA investment banker, but I've seen them work on this stuff a bit.

But it's how a huge amount of financial business gets done. Think of it as a powerful multidimensional calculator - an engineer would not resort to a program written by IT every time they want to work out the forces acting on a girder - in fact a former coworker used to use a spreadsheet to compute the voltages and currents in electronic circuits (I assume he still does). Spreadsheets are just visual programming tools - a set of mostly-canned programs that you can plug values into, and combine together - and can be used, and abused.

Modeling is primarily about trying things to see what you get, and then trying other things to confirm from another angle. Often you don't know what the angles are until you are doing it.

See Excel Shortcuts: A Modeling Must

And this from "Things investment bankers love":

If this list were in any type of order, Microsoft Excel would probably be at the top. This brainchild of Bill Gates allows investment bankers to do things our parents could have only imagined. This program allows us to value companies using all types of inputs, assumptions and formulas. Back in the old days, bankers had to do this on pen and paper, and most valuations were just like Ebay auctions with bidders using their gut to value the company. And if you decided you wanted to change one little number in the 1,000 rows of data? Well, basically you were screwed. But with handy Excel, once you have the setup in place you can make small changes and everything flows through. It’s simply amazing. Microsoft Excel allows me to do anything my bosses’ hearts desire. Want to see an accretion/dilution analysis by 9am tomorrow morning? No problem. Excel and I will get to work! It truly is amazing to think how things used to be done.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

When it has 3 90deg angles being on the surface of a sphere.

That is still a 2D surface as you only need two numbers to specify any given point on the surface.

Re:If you need it you are doing it wrong.

[TemporalBeing]

Sure, but I think the idea is that if your spreadsheet doesn't do all the calculations it needs to do more or less instantly, on any computer built within the last ten years, you're either a) horribly misusing your spreadsheet, or b) using a horribly written spreadsheet program.

I wouldn't necessarily agree.

For instance, I used a spreadsheet to do an algorithm comparison. For the comparison I had to have thousands of rows of data that were calculated with dependencies on previous rows with multiple sets and multiple spreadsheets doing this (e.g. 5 columns made a set, 5 sets in one spreadsheet, and again in another, then additional spreadsheets utilizing those to do certain tests, again for the same entire data set, for each algorithm I was comparing - in this case just two algorithms). It takes Calc a few minutes to do the calculations - and I would expect it to.

Now, Calc loads slower than the equivalent XLS/XLSX document for the reason I mentioned earlier. However the file size is dramatically different. The ODF version is only a couple megabytes at most (well under 5...possibly under 1). The XLS version is 23 megabytes. Getting Excel to recalculate it isn't much different in time - it's just the load time that differs.

Now, you could argue that a program would likely be better than a spreadsheet; but it would also take a lot more to get to doing the comparison too - having to build an entire GUI for something just to be able to manipulate the values and see the differences. So for a test, it worked great and did what I needed it to do.

It's kinda like redesigning your VW Bug to be able to back up to a full-on commercial loading dock to load groceries.

Now that said, sure, why the hell not? The GPU is just sitting there, might as well make use of it. But it's more along the lines of 'why the hell not' than 'this will make it faster!'

As in my case, I doubt using the GPU would really speed it up all that much. Sure, you could probably hand each data set row by row to different GPUs, and get some benefits of parallel processing - but it'd be a real PITA for it get it right.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

You also need to define the size of the sphere, so that makes it 3D.

Re:If you need it you are doing it wrong.

[brantondaveperson]

far and beyond the point where it is reasonable to switch to a more powerful tool

Nonsense. Spreadsheets do a great job of all sorts of things. Why shouldn't they be made more powerful? A spreadsheet is a database, albeit a slightly specialised one. Why shouldn't I be able to drop 100,000 rows of data into it and have it calculate/plot whatever it is I'm interested in? How will a 'database' do this any better?

switching over to using something like MS access

Crikey. I'll assume you're joking.

Now, what spreadsheets do need is a better UI. I have gigs of ram and megagiggawatts of CPU power and 8 cores and all the rest of it. I should be able to zoom in & out of enormous spreadheets containing gigbytes of data without having my machine grind to a halt. If you need the GPU for this, more power to you. Just make it fast. Have a view that shows all the dependencies between cells, have the ability to specify that these cells are a matrix, and operate on that matrix accordingly. Implement better graphing tools (excel, I'm looking at you here...). Etc etc.

Maybe such a product would have a different name? Sheetbase? Dataspread?

Re:If you need it you are doing it wrong.

[NemoinSpace]

Throwing more hardware to compensate for bad code has been the practice for,uh,ever.
I appreciate good code as much as the next guy, but programmers are expensive, and good programmers are kinda weird. Besides, there is plenty of cloud left.

Re:If you need it you are doing it wrong.

[SuiteSisterMary]

Quite the edge case, as you'll agree, I'm sure. But I think we'll also agree that the effort going into this new feature, which will affect few people in any meaningful way, could probably be better put towards optimizing the base software, which would benefit many, if not all, users.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

Not at all, but I will flat out tell you that I've seen domain specific spreadsheets which have surprisingly little to do with adding numbers

Yup. Like the floor plan of an office of 2000 people. Each desk a drawn by a group of cells with a grey border the sheet itself zoomed out to 5% viewing area. Red dot for those desks with UPS, and ethernet ports a blue spot.

Excel is often the only tool flexible enough for the job, in corporate locked down environments and elsewhere. Better Excel than MS Paint.

Re:If you need it you are doing it wrong.

[cheesybagel]

Yeah. I mean dear god please no. When you crash a GPU program you usually hose the entire operating system and need to do a hard reset. GPU programs run in a kernel mode driver. Blech.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

The problem is that unlike MS Office, the ODF format does not typically store pre-calculated values of formals; so when you load a file it has to run all the formulas, etc to generate what the user wants to see. There's nothing wrong with that, it's just a little slower. An easy optimization would be for ODF to store formula results as part of the cell contents - but that's a change to the ODF standard AND the software that uses it.

 

That is just wrong. ODF also writes cached values into the file format and Libreoffice can use the cached values since Version 4.0. There are still some minor issues because nobody ever used the cached values in ODF before but beginning with Version 4.1 it will just work and you won't notice that the document has not been calculated during import.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

A database can't solve the problem GPU-acceleration solves for the spreadsheets.

Re:If you need it you are doing it wrong.

[sjames]

Well, that certainly explains the state of our economy.

Re:If you need it you are doing it wrong.

[PiMuNu]

The problem with excel is that it is hard to document against and test. In the software world we have lovely tools like doxygen and all sorts of testing suites. We have nice things like functions, classes. One can do analogous things in spreadsheets, but they just aren't as well set up for that sort of stuff. It makes it hard to define what is going on into logical blocks, hard to break that down into blocks that can be tested against. I'm not someone who cares about this language or that language, but I do care that a developer goes through some development process that includes testing and documentation, and I don't see how that can be done in excel.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

No it doesn't. Here I thought slashdotters were nerds and actually knew one or two things about mathematics.

Re:If you need it you are doing it wrong.

[Anonymous Coward]

As punishment for your ignorance, you will be sent to the supplies department where you will tell them you have "come for a long stand".

Re: If you need it you are doing it wrong.

[Anonymous Coward]

While I do agree it would benefit the edge cases like mine above, it will also benefit all users for the simple fact that I pointed out earlier that ODF doesn't store the calculated values of formulas, so even minor spreadsheets would have to redo calculations at load - calculations that may be far more parallizable than my example.

TemporalBeing - posted Anonymously as not logged in

Re:If you need it you are doing it wrong.

[Grismar]

You could also argue that, if you need a modern GPU to be able to render the graphics for a fun game, you're doing it wrong! What's wrong with Pacman and Space Invaders, right?

More space allowing for more data and faster calculation in a spreadsheet allows for more uses. Just the fact that you'd pick another set of tools, doesn't mean someone able to do it quickly and effectively with a spreadsheet is doing anything "wrong".

Re:If you need it you are doing it wrong.

[Anonymous Coward]

Also MS Excel is seriously fast considering what it does. They have a lot of optimizations to make it recalculate the way that it does. I think MS software is generally quite poor but Excel is really a stand out in quality. I write hugely complex spreadsheets in Excel and also know how I would write the same in traditional code. The spreadsheets are not so far behind.

Re:If you need it you are doing it wrong.

[hawkfish]

A good example of this approach is columnar databases (C-Store, MonetDB etc.). The performance gains over traditional implementations are generally cited as 200x (which is about what I see in the one we built.) The gains are largely from simply changing the data structures:

* Columnar decomposition (i.e. big honking memory mapped C-style arrays) reduces the amount of disk IO
* Columnar representation allows you to trade disk IO for lightweight decompression in the CPU
* Lightweight compression techniques make it easy to operate directly on the compressed data
* Columnar representations also allow you to stop building tuples every time you compute a new attribute ("late tuple assembly")
* Working on data blocks that fit in L1 allows you to fit the computation in the L1 instruction cache ("block iteration")

By contrast, a system like MySQL uses an interpreter to process rows one at a time by using this traditional "accessors" approach, and it is dog slow.

Re:If you need it you are doing it wrong.

[lsatenstein]

Thats not the issue. If your spreadsheet is SO larger that on a MODERN CPU, its slow ... you're doing it wrong.

You can make insanely complex, application like spreadsheets, without noticing 'recalc' time. By the time you get to noticing 'recalc' time, you've fucked up.

Caveat: OO.org is known to have some of the crappiest code in existence, so with the case of Calc, you don't have to make ridiculous spreadsheets to notice recalc time. GPU support won't fix the problem however as its not the math thats the issue, its the shitty logic code filled with stupid crap written by clueless devs that cause Calc to be so slow.

===
My experience with larger sheets was with references from one sheet to another. It is that occasionally, libreoffice software has to wait for garbage collection to free up some fragmented working memory. A few seconds of pause and then its again off to the races. So, going the GPU route may help, and perhaps better software memory management would be a priority problem to tackle.

Re:If you need it you are doing it wrong.

[badkarmadayaccount]

Yes.

Re:If you need it you are doing it wrong.

[aztracker1]

Sorry, it isn't really OO vs. Functional in this instance so much as the mindset of what I've seen as a typical OO programmer.

Clarification

[UnknowingFool]

From the article:

Calc is based on object oriented design from 20 years ago when developers thought that a cell should be an object and that creates a huge number of problems around doing things efficiently.

The problem isn't that Calc is object-oriented but was designed such that many things depended on the spreadsheet cell.

Re:Clarification

[Trepidity]

Yeah, and it sounds like the GPU angle is really just a hook to get AMD funding. The more important improvements will be refactoring the representation so it doesn't suck in the first place.

Re:Clarification

[should_be_linear]

Cell should be an object even today. Their problem is probably, that Cell object contains something like string object, so creating 1 million of cells meeds million pointers and allocations to million of strings, which is performance killer. What they need to do is: instead of string, put int handler of string into cell, and have all strings in single huge allocated blob (like: StringBlobMap object). Going away from objects to improve performance is rarely good idea.

Re:Clarification

[robthebloke]

No. Just no. Wrapping a float in it's own object is completely retarded. The performance of float + float style computation hasn't really improved that much in the last 7 or 8 years. So if you want to improve the performance of your app, you're going to need to parallelize it. The easiest way to achieve this is to stop thinking about an object that represents a float, and start thinking about an object that represents an array of floats. Typically, it's far easier to optimise simple arrays for SIMD & multi-core, rather than lots of small objects. Fine granularity OOP has been dying for a few years, but it's now so far beyond dead, that no one should be advocating that approach in any software product.

Re:Clarification

[Mr Z]

I think it's probably fine to store non-empty cells as objects, as long as you use something like the Flyweight pattern to avoid carrying too much baggage in each cell. It makes for a fine user interface, I'm sure.

To really get good recalc performance, though, you really need to drag the cell dependence graph out of the grid and treat it like an actual program. Once you do that, you could actually JIT the computation represented by all those cells. If you're really walking the object forest for every recalc, you'll never speed up.

Re:Clarification

[tyrione]

From the article:

Calc is based on object oriented design from 20 years ago when developers thought that a cell should be an object and that creates a huge number of problems around doing things efficiently.

The problem isn't that Calc is object-oriented but was designed such that many things depended on the spreadsheet cell.

OOA/OOD isn't the problem. It's the team of architects who came up with their design using OOA/OOD that is the problem. Lighthouse Design using NeXT's MVC paradigms had Quantrix and Parasheet, Diagram!, Tasmaster, Concurrence [precursor to Keynote] and more. Some of the architects are at Apple and have been working on iWorks.

Refactor? APU?

[JBMcB]

If the refactor is done properly I don't think the OpenCL acceleration would be necessary. Heck, 1-2-3 running on a 486 was pretty speedy.

Re:Refactor? APU?

[Capt.DrumkenBum]

486, Bah humbug.
I ran Lotus Symphony on a 286. It was the job that convinced me to go back to school for a career in IT.
Also, get off my lawn.

Re:Refactor? APU?

[Jah-Wren Ryel]

If the refactor is done properly I don't think the OpenCL acceleration would be necessary.

They are going to need it for the flight simulator function.

Re:Refactor? APU?

[Valdrax]

A 286? Luxury!
I used to use AppleWorks on an Apple II GS running a 65C816 chip, and that's the system that convinced me that programming was a fun hobby.

Re:Refactor? APU?

[TeXMaster]

While OpenCL might not be necessary, there's no reason not to use it, since it will mean easy, cross-platform support for multicore programming and use of vector functions, that would be useful on any modern system, even just on CPUs. (Of course, if the system also has an APU with the upcoming hUMA architecture and can access the same memory space as the CPU, why not make use of it too?)

Of course the benefits will only be visible to people with huge spreadsheets. As for 1-2-3- being speedy on a 486, well, either you only had small spreadsheets or the mist of the past is obscuring your memory.

Re:Refactor? APU?

[Capt.DrumkenBum]

The machine that convinced me programming was a fum hobby was a Timex Sinclair 1000, with a Zilog Z80A @ 3.25 MHz, and 2k of RAM

Re:Refactor? APU?

[Anonymous Coward]

Actually, there was a spreadsheet on the Nascom ][ (Z80 processor, 8k ROM Basic, about 48kB of RAM) that converted its cells into Basic statements it fudged into the ROM Basic memory. Recalculation then just ran the program and siphoned the variables back onto screen. I think the whole spreadsheet program was about 1kB or so. Yes, I got my units right. And it was pretty snappy at 4MHz CPU speed.

Re:Refactor? APU?

[jeremyp]

Visicalc on the Commodore Pet.

The 6502 was a 1MHz machine and floating point operations and even integer multiplication had to be done in software. How can a spreadsheet program designed for CPUs more than 2000 times faster with specialist floating point hardware be too slow? It must be shitty code.

Re:Refactor? APU?

[cnettel]

Visicalc on the Commodore Pet.

The 6502 was a 1MHz machine and floating point operations and even integer multiplication had to be done in software. How can a spreadsheet program designed for CPUs more than 2000 times faster with specialist floating point hardware be too slow? It must be shitty code.

It wouldn't be unlikely that there are more than 100,000 times as much data. Add to that automatic recalculation and graphic rendering. That said, I still consider the current performance tragic.

Re:Refactor? APU?

[Lord Crc]

If the refactor is done properly I don't think the OpenCL acceleration would be necessary.

Did you miss the APU part there? The future, at least according to AMD, is a weaksauce CPU glued to powerful compute units.

So using OpenCL for Calc would be similar to using the hardware video accelerator to make that i3 play back 1080p h264 content without sweating.

Re:Refactor? APU?

[Mr Z]

For me, it was AppleWorks Spreadsheet on an Apple ][e. Slightly newer than a Commodore PET, but the same 1MHz 6502 goodness. And then, SMART Spreadsheet on a 6MHz 286 AT clone. (SMART 1.x, not SmartWare II. SmartWare is still around, believe it or not.)

Re:Refactor? APU?

[tyrione]

If the refactor is done properly I don't think the OpenCL acceleration would be necessary. Heck, 1-2-3 running on a 486 was pretty speedy.

I'd love to tie in R/Octave and do Numerical Analysis with Calc, so that OpenCL would be very handy indeed, seeing as it uses the CPU/GPGPUs and any other processor/co-processor like DSPs.

Non-linear solver engine

[Anonymous Coward]

Will it finally have a non-linear solver engine for double-checking code output against a spreadsheet??

Huh?

[Anonymous Coward]

To be more performance? Good to see the "editors" are as incompetent as always.

Re:Huh?

[Cyko_01]

Good to see the "editors" are as incompetence as always.

There, FTFY

Re:Huh?

[Anonymous Coward]

Slashdot has editors?

the problem with OpenOffice

[Anonymous Coward]

There has to be a substantial, paid, fulltime project team to do UI, feature work, localization, QA and regression testing, doc, and document interoperability/backwards compatibility for each release. Testing has to cover many OS releases, video drivers, and targeted display devices as well as end user natural language, and include reduced RAM/swap space installations. That's what Microsoft has. That's why I don't consider OpenOffice a bargain even for free.

Re:the problem with OpenOffice

[Bert64]

And despite all this, and the fact they target far fewer platforms, MS still puts out an extremely buggy application, with poor forwards/backwards compatibility, virtually no intentional interoperability with anything else, and major inconsistencies between the 2 platforms that they do support.

The only difference with MS bugs is that users have come to expect them and know the workarounds. Ask anyone who uses ms applications heavily and they will have all kinds of kludgy hacks memorised to get around functionality which doesn't work correctly.

Re:the problem with OpenOffice

[Anonymous Coward]

How does office have poor forwards and backwards compatibility? Troll much?

Re:the problem with OpenOffice

[dugancent]

Poor backward and forwards compatibility? Really?

I recently opened a few documents with Word 2011 on my Mac that were authored with whatever version of Word that was available on Windows 3.1/3.11. After a quick conversion, they loaded perfectly. I regularly send and receive documents with a friend who is still running Office 2003 on XP, again, no problems at all.

Re:the problem with OpenOffice

[Anonymous Coward]

Troll yourself. try open an Office 5 document with your shiny Office 360. Good luck with that.

Re:the problem with OpenOffice

[gl4ss]

How does office have poor forwards and backwards compatibility? Troll much?

well in the usual way that you need to keep overlapping versions around.

though, the anon gp is putting a wee bit too much on how much ms puts effort into their things... ms doesn't run tests on 256mbyte ram hw just for seeing that there's not enough memory. I seriously doubt ms even tests different video drivers since they have little to do with office itself(and if the driver passed certification it should run it). on arm side ms just dictated what hw it can run on so not that much testing to be done there either(except that using it on the touchscreen sucks).

Re:the problem with OpenOffice

[Bert64]

It's well documented, you can find examples all over google, eg:

http://hints.macworld.com/article.php?story=20111230095628470

Infact there are many people who use libreoffice to open and convert corrupted (or very old) files which are making msoffice crash, libreoffice is far more tolerant of unexpected data in the input files as unexpected data is a given when attempting to reverse engineer undocumented formats.

And to give one personal example, msoffice 97 onwards had a bug in the macro function whereby the line counting function ignored lines with bullet points, so we had an extremely kludgy macro which counted the lines and then iterated through looking for bullet points and increased the line count accordingly... MS decided to fix this particular bug in a "security update" for office 2003, but then reintroduce the bug in 2007... Obviously this kludgy macro catastrophically broke the day that patch got rolled out.
I could understand if it broke going from 2003 to 2007, but not for what is supposed for be a security update to change something like that.

Also even moving files between the exact same patch release of msoffice on different machines can cause problems with formatting, as it reformats depending on available fonts and printer settings.

Re:the problem with OpenOffice

[jones_supa]

There has to be a substantial, paid, fulltime project team to do UI, feature work, localization, QA and regression testing, doc, and document interoperability/backwards compatibility for each release. Testing has to cover many OS releases, video drivers, and targeted display devices as well as end user natural language, and include reduced RAM/swap space installations. That's what Microsoft has. That's why I don't consider OpenOffice a bargain even for free.

Wise words, my brother. That's exactly the problem with many OSS projects. Modern software is so complex that you really want solid funding and good bunch of full-time developers and quality assurance people to make a perfect product. If this was the case with OpenOffice too, it would have already steamrolled MS Office.

Re:the problem with OpenOffice

[ledow]

Publisher. For YEARS. Maybe the very new versions are different, but I have this every day with people bringing in old/new versions of Publisher files.

And no import filter is ever perfect, even if it was written by MS, for opening MS formats, for a MS application.

Re:the problem with OpenOffice

[mdielmann]

Poor backward and forwards compatibility? Really?

I recently opened a few documents with Word 2011 on my Mac that were authored with whatever version of Word that was available on Windows 3.1/3.11. After a quick conversion, they loaded perfectly. I regularly send and receive documents with a friend who is still running Office 2003 on XP, again, no problems at all.

Without an idea of what is in the files, it's hard to say how relevant this is. You can effectively use Word as Notepad with a nicer interface. You can also use Word with piles of macros, multi-level formatting, custom styles, and that weird font you downloaded 8 years ago. One will be likelier to convert from one version to another than the other will.

I'm not saying that you're presenting a specious argument, but one of the tricky things in this discussion is determining which feature MS decided to update between one version of Word and another, and whether that will cause a conversion to break. But if I was going to make a work-alike of Word, I'd take special note of the features that tend to fail on conversion, and pay extra attention to them when I make my conversion routines. After all, if the producer of both versions of Word has trouble with conversions between them, I should take that as an indicator that it requires special effort. And it looks good for me.

Spreadsheet applications shouldn't need GPU!!

[MetalliQaZ]

Am I the only one that notices how crazy that sounds?

Re:Spreadsheet applications shouldn't need GPU!!

[robmv]

OpenCL doesn't mean it will need a GPU but that it can use one if available. OpenCL can use your CPU and there will be performance advantage on those cases too, they can use tuned OpenCL libraries instead of rewriting everything inside LibreOffice

Re:Spreadsheet applications shouldn't need GPU!!

[camperdave]

On the other hand, massively parallel computation is what spreadsheets are and what GPU is good at.

Re:Spreadsheet applications shouldn't need GPU!!

[tech.kyle]

Sure, rewriting the code to be more efficient and then throwing in OpenCL acceleration IS a bit of a redneck approach to things, but at the end of the day, why wouldn't you want the metaphorical V10 crammed in to your spreadsheet software? If it can be done gracefully, why not?

Worth mentioning that Good Guy AMD isn't making it proprietary.

Re:Spreadsheet applications shouldn't need GPU!!

[wonkey_monkey]

No-one's saying spreadsheets need GPU acceleration. But why shouldn't the GPU be taken advantage of?

Re:Spreadsheet applications shouldn't need GPU!!

[gl4ss]

Sure, rewriting the code to be more efficient and then throwing in OpenCL acceleration IS a bit of a redneck approach to things, but at the end of the day, why wouldn't you want the metaphorical V10 crammed in to your spreadsheet software? If it can be done gracefully, why not?
Worth mentioning that Good Guy AMD isn't making it proprietary.

well, given their track record with performance it would be better if they first sorted out the performance bottlenecks elsewhere and then brought in opencl for doing raytrace calculations from the spreadsheet.

Re:Spreadsheet applications shouldn't need GPU!!

[Anonymous Coward]

Hey, it boosts the performance of the built-in FlightSim easter egg.

Oh wait, that was Excel.

Re:Spreadsheet applications shouldn't need GPU!!

[corando]

On the other hand, massively parallel computation is what spreadsheets are and what GPU is good at.

^^^ This. Although not always, but often enough.

If you have multiple columns performing the same action for each cell accost a few thousand rows, it could make a difference.

Re:How is it?

[MightyYar]

I don't think most people say Calc is just as good as Excel - they say that it is good enough for most people. And that is probably true. I think my boss uses excel for simple formulas and for lists. I use Excel for anything not quite worthy of a Matlab script, so OpenOffice doesn't quite measure up for me but should work fine for my boss.

Libre Office Calc isn't that good.

[Anonymous Coward]

Engineers use spreadsheets for data analysis, data scientists, you name it - and of course accountants.

Why?

Because when you are doing a very specific task spreadsheets are the best tool. Having to code a Python or Perl script isn't worth it and to do what a spreadsheet was designed to do requires a bunch of libraries with their own API. Why go through all that coding when a spreadsheet can do it. I am doing one off ananlysis and to write programs that I'm gonna never use again - spending all that time debugging, testing, etc .... for something that I'm going to use once? Forget it.

I know most accountants love the spreadsheet and they make insanely complicated things using spreadsheets pushing it far beyond what these are designed to do.

If they weren't designed to do things then they wouldn't be able to do them, would they?

I do a lot of analysis and I've used various spreadsheet programs. I work mostly on Linux and I use Gnumeric - it's MUCH better than Libre Calc. Especially when you're importing CSV files. Libra calc just shows some Asian characters.

The Gold standard is MS Excel and if MS had a Linux version of that, I'd be on it like a sailor on a $2 whore who's been out to sea for a year - the sailor who's been out to sea - not the whore.

I don't want to say Libre Office Calc sucks, but it's not the best spreadsheet out there.

Re:Libre Office Calc isn't that good.

[Bert64]

Gnumeric is great, tends to focus on accuracy too whereas msexcel truncates numbers over a certain size (among other weirdness) and libreoffice seems to copy it...

However...

If they weren't designed to do things then they wouldn't be able to do them, would they?

I could use a 40 ton truck to deliver a single letter, i could drive myself to work in that same truck, and if all i had was a 40 ton truck and i only wanted to do the journey once or twice i might even do so, despite how awkward it would be driving such a large vehicle not to mention the fuel usage.

On the other hand, the more often i did these things the more inefficient it would be to use a 40 ton truck for such things, and i might be better off buying a regular car or even a bike.

An unsuitable tool might do as a temporary substitute, but long term you really want to use an appropriate tool for the job.

Re:Libre Office Calc isn't that good.

[gstoddart]

An unsuitable tool might do as a temporary substitute, but long term you really want to use an appropriate tool for the job.

Look at it this way ... the 40-ton truck in your metaphor (Excel or something like it) is provided to everyone in the company from day 1. From the receptionist to the CEO, everyone gets a 40-ton truck. You know that everyone can carry the same stuff in their 40-ton trucks because they are all pretty much the same.

Furthermore, before you even leave highschool, people tech you how to use that 40-ton truck.

Now, imagine that you need to solve a new problem, which is shockingly similar to problems you've already solved.

So you could go through 6 months to a year of fighting to get someone to help you build a station wagon with a baby seat and tinted windows, because the 40-ton truck is overkill. And you need to convince someone help pay for the station wagon since they didn't budget for one of those.

After you've gone through all of that process, the station wagon has never materialized, the cost overruns make it look like you're buying a gold-plated Rolls Royce, but the engine is still a cardboard mock-up, and the people building it for you have forgotten to include headlamps, windshield wipers, turn signals, seatbelts, and a speedometer. But if you will submit a change order to have them build those, you can wait another period of time (and even more money).

Or, you take the 40-ton truck to do what you need, take a little extra time to find a parking spot, and in the end you've got something which covered your needs in a shorter period of time and for no extra costs except your time. You can get to the grocery store and back in a few hours, and you're done.

That is why people use spreadsheets and don't always jump straight for the custom application.

Re:Libre Office Calc isn't that good.

[Anonymous Coward]

I have done a lot of my power budget number on excel i.e. what power rails each of the chips need. I give the thermal design people the spreadsheet so that they can put the power dissipation of the chips into their thermal model. I give the power guy the same spreadsheet so that he can design power supplies to handle the chips.
I give the FPGA team the same spreadsheet for review on the amount of chip usage/frequency power estimation. I can easily do charts (e.g. percentage of power consumption or power vs chip frequency) a few mouse click away, Heck even my manager asked to play around with my spreadsheet.

For each of those groups, we can all share the same info and not having to install a compile environment so that they can update/change the parameters. At the end of the day, I can copy & paste the final spreadsheet into the design document as is.

I don't see how a user written program can do that nor having to train everyone in the diverse team without a major fight on the language/environment/"framework" of choice.

Re:Libre Office Calc isn't that good.

[Zero__Kelvin]

How did you know I use spreadsheets and drive a 40 Ton truck !!!???

Re:Libre Office Calc isn't that good.

[Zero__Kelvin]

"Heck even my manager asked to play around with my spreadsheet."

You should go to Human Resources right away. That's sexual harassment!

Re:Libre Office Calc isn't that good.

[gstoddart]

How did you know I use spreadsheets and drive a 40 Ton truck !!!???

Doesn't everybody? ;-)

Re:Libre Office Calc isn't that good.

[Anonymous Coward]

yep, working in high powered led lighting we do that too, it doesnt take that long to setup, and its very easy to take a "I want X lumens" and see how much current is needed and how much heat is generated from dozens of competing products to find the best match.

Why not Javelin?

[jabberw0k]

Spreadsheets lead the inexperienced, down the garden path of "Oh, this looks easy..."

At some point you think, Oh, let me just sort this column. And you fail to realize some formula on sheet 27 presumes a linkage between column C on sheet 5 and column F on sheet 13. So now your entire model is garbage.

In all these decades, hasn't anyone resuscitated Javelin with its time-oriented models, where what looked like a spreadsheet was just a view of the underlying model? "Javelin understands the arrow of time" -- 1985 slogan

Re:Why not Javelin?

[Anonymous Coward]

Spreadsheets lead the inexperienced, down the garden path of "Oh, this looks easy..."

At some point you think, Oh, let me just sort this column. And you fail to realize some formula on sheet 27 presumes a linkage between column C on sheet 5 and column F on sheet 13. So now your entire model is garbage.

In all these decades, hasn't anyone resuscitated Javelin with its time-oriented models, where what looked like a spreadsheet was just a view of the underlying model? "Javelin understands the arrow of time" -- 1985 slogan

because not everyone uses an spreadsheets as accounting sw?

Re:Why not Javelin?

[WillAdams]

Agreed. It kills me that one has to shell out $1,495 to get Quantrix Financial Modeler to have that sort of interface w/ the numbers, and it's still bizarre to me that Lotus Improv didn't succeed.

Re:Why not Javelin?

[MiG82au]

Then you're (not you personally) an idiot that's ignored Excel's suggestion of "there is adjacent data. Would you like to extend your selection?".
Seriously, I figured out sorting back in school when I started using spreadsheets, so it doesn't take that much experience.

Sick of bloody programmers getting all high and mighty about how I should do engineering.

and still no "normal view"

[cellocgw]

Why oh why can't the bleeptards at LibreOffice recognize that proper document editing is done in a "Galley View" which MsoftWord refers to as "Draft" (previously "Normal" ) view? Displaying page boundaries, headers & footers, etc is of exactly zero benefit while one is composing the text of the document.
Personally, I'd like not to see text formatting either (bold, font size, etc) but I can live with that. At least until I find a company that supports LaTex, anyway. For that matter, why couldn't LibreOffice (and Micrsoft too) have a twin-pane editor like TexMaker? Do your typing in one pane and observe the fully rendered page in the other as desired?

grrrrrrumble

Re:and still no "normal view"

[EvanED]

Displaying page boundaries, headers & footers, etc is of exactly zero benefit while one is composing the text of the document.

I beg to differ, sir!

It is not exactly zero benefit. It is actively distracting, and hence of negative benefit.

Incidentally, there's an 11-year-old bug report with 281 votes (there are only two bug reports with more votes) if you want to add your voice. I rarely use word processors (Latex here, as much as I hate it there's not really anything better for what I need), but if I did, I'd use Word almost on account of the lack of a normal view alone.

Re:and still no "normal view"

[WheezyJoe]

Why oh why can't the bleeptards at LibreOffice recognize that proper document editing is done in a "Galley View" which MsoftWord refers to as "Draft" (previously "Normal" ) view? Displaying page boundaries, headers & footers, etc is of exactly zero benefit while one is composing the text of the document. Personally, I'd like not to see text formatting either (bold, font size, etc) but I can live with that. At least until I find a company that supports LaTex, anyway. For that matter, why couldn't LibreOffice (and Micrsoft too) have a twin-pane editor like TexMaker? Do your typing in one pane and observe the fully rendered page in the other as desired?

grrrrrrumble

C'est a little off topic, but I so very much agree. Top reason I can't cut the M$Word cord for Writer. Please, LibreOffice people, please?

Re:and still no "normal view"

[Bert64]

LibreOffice refers to this as "web layout", and its right there in the view menu.

As for why it doesn't work like LaTeX, i guess thats because its aiming at a totally different market... Most people simply don't understand the idea of formatting being separated from content, they just want to lay the page out as they see it - as if they were doing it by hand. Also modern word processors have moved more towards traditional DTP applications, where there is a focus on layout rather than on typesetting a large body of text.

Re:and still no "normal view"

[CastrTroy]

This is why I loved WordPerfect 5.1 so much. Because of the simple text based interface, you didn't spend so much time worrying about how your document looked, and just spent time typing up the actual document. All the features were available from the keyboard which meant that it was faster to do any kind of formatting that you needed to do because you never had to move your hands away from the keyboard.

Re:and still no "normal view"

[EvanED]

LibreOffice refers to this as "web layout", and its right there in the view menu.

No. LibreOffice's "web layout" is the same as Word's "web layout", which is different from Word's draft/normal view. And (at least IMO) it's even worse than page layout.

Page layout displays the page to too great of fidelity, because the top and bottom margins break up the flow of text. Web layout goes too far in the other direction, completely reflowing the text to the width of the window.

Re:and still no "normal view"

[jkflying]

My version of LibreOffice has this (4.0.3.3).

Menu --> View --> Print Layout toggles the behaviour.

Re:and still no "normal view"

[Anonymous Coward]

Agreed. It should be possible to start the editor in a simple, clean, typesetter-concern-free mode that has as little clutter and distraction possible. That the default mode looks like something only a SolidWorks user would appreciate is a giant fail.

Re:and still no "normal view"

[EvanED]

...reflowing the text to the width of the window

And to finish my thought, the reason this is bad is because the text area is then too wide unless your window is too narrow to contain full toolbars.

Re:and still no "normal view"

[EvanED]

My version of LibreOffice has [an equivalent to normal mode] (4.0.3.3).

No it doesn't.

Re:and still no "normal view"

[cellocgw]

Yeah, that bug was opened by yours truly.

And I do agree that headers are of negative (not zero) benefit

Re:and still no "normal view"

[jendral_hxr]

How about some plain text editor? vi/vim? nano?

Thats the OOP mentality

[Anonymous Coward]

From the article:

Calc is based on object oriented design from 20 years ago when developers thought that a cell should be an object and that creates a huge number of problems around doing things efficiently.

The problem isn't that Calc is object-oriented but was designed such that many things depended on the spreadsheet cell.

The OOP mentality is that you make everything an object. And when you have developers who drank the OOP Kool-Aid, you get problems like this.

Re:Thats the OOP mentality

[UnknowingFool]

And what other model should they have used? Procedural? Assembly? The approach that was picked 20 years ago needs to be revised. It isn't about whether or not to use OOP.

Re:Thats the OOP mentality

[YoungManKlaus]

wrong. the problem is that you can make a bad design based on every concept. If you have your base objects small and fast its no problem, but if your cell mutates already in an abomination of hundreds of lines then something is quite wrong, regardless if its a struct with helper functions or an object ;)

QA Testing

[Anonymous Coward]

This sounds neat, but you know what's a lot more important to me: quality assurance and testing. I use Calc for a variety of things, and just this past week tried using LibreOffice but I had to switch back to OpenOffice because LibreOffice was riddled with bugs. Most egregiously, after saving documents in LibreOffice I could not open then in OO. I also noticed that a lot of formatting setting were not preserved that worked in OO. And this was after only spending a few hours playing arond. I'm sticking with OpenOffice.

Re:QA Testing

[jones_supa]

That is the worst problem with open source these days. Little bugs and misfunctioning parts everywhere. Quality assurance is needed more than anything.

Re:QA Testing

[Zero__Kelvin]

How do you know that it isn't Open Office that has the bugs? (Actually, they each likely share part of the blame)

Slow?

[Anonymous Coward]

What I find interesting about this is the assumption Calc is slow. I have used OpenOffice and LibreOffice for around a decade and performance has never been an issue. Granted, my spreadsheets aren't huge, they're mostly just columns and 2-D tables, but even on lower-end hardware I've never noticed a performance issue. I think it's great that they are looking at speeding up the application, but I don't see it as something which needs doing.

Proper analogue for Visio next?

[Anonymous Coward]

Tangential, but... LibreOffice "Draw" is not a good substitute for Visio. It barely does the most basic functions. Can AMD fund them creating a _good_ alternative to Visio please?

Back on topic - I've seen some issues with Calc. Most of the time it does most of what I need, but usually does it with different names for the function calls. Hopefully a good refactoring will make it easier to add/improve features.

Re:Proper analogue for Visio next?

[tibit]

It's only recent that Visio format is supported at all. Give them some time. It's a lot of work. Note that LibreOffice is, at the moment, IIRC the only piece of software that can open the early Visio 1 and 2 format files. Not even Microsoft Visio can open them (feel free to correct me if I'm wrong, of course).

Right (Re:If you need it you are doing it wrong.)

[WillAdams]

Actually, the UI for Lotus Improv was quite nice and won some awards.

Its (spiritual) successor, Quantrix Financial Modeler seems to be selling well enough, even w/ a $1,495 price point.

I wish that Flexisheet (an opensource take on this sort of thing) would get more traction.

Live a day in my shoes

[sjbe]

If your spreadsheet needs a gpu to speed up calculations, you are probably misusing spreadsheets.

Or it just means that you have some pretty complicated calculations. More computing horsepower never hurts.

I know most accountants love the spreadsheet and they make insanely complicated things using spreadsheets pushing it far beyond what these are designed to do.

I happen to be an accountant as well as an engineer. What pray tell do you think spreadsheets were designed to do? (hint - it involves rapid data modeling) They aren't much use if the only problems you solve are toy problems. Plus they require relatively little training to use effectively. Someone can be trained to solve real world problems MUCH easier than with most other tools. Most of the problems I'm asked to solve are ad-hoc investigations into specific questions. I shouldn't need a four year degree on Comp-Sci to accomplish a bit of data modeling.

But if you have a spreadsheet that needs this much of cpu time to recompute, you should probably be using a full fledged data base with multiple precomputed indexing.

I use some rather complicated spreadsheets. A database would be of no advantage whatsoever for 99.9% of what I use a spreadsheet for. Furthermore a database would be a lot slower to develop, harder to update, and require significant user interface development. If I'm crunching sales data or generating financial projections a spreadsheet is almost always the easiest and most useful tool for the job.

Databases come into the picture when: A) other applications need to interface with the data, B) the dataset becomes truly enormous, or C) the number of dimensions in the data exceeds 2 to 3. Sometimes I use databases. Most of the time they would be a waste of money, brains and time. Frequently when I actually need a database I'll create a mock up of the tables and calculations on a spreadsheet first which lets me work out the structure much more easily.

While it is certainly possible to use a spreadsheet inappropriately, a spreadsheet should be able to handle a rather large amount of data and calculations before it chokes.

Re:Live a day in my shoes

[tibit]

You should know basic programming upon exiting high school. Yeah, the sad state of K-12 curricula are something to lament another time.

So, what tool do you use to diff your spreadsheets? How do you ensure that there isn't a bug in a column of otherwise "identical" formulas? How do you ensure that whatever column you've filled with imported data still has this imported data in it? Where's your log that shows that you haven't unlocked some cells by mistake and messed them up "subtly"?

Spreadsheets provide a semblance of productivity and an illusion of efficiency. Once you go from fucking about with it to being serious, spreadsheets are a gift from hell. It takes very little to audit a couple pages of code. Good luck auditing a spreadsheet that has tens of thousands of cells filled with formulas.

Once you use spreadsheets in an auditable and professional way, you end up with a VBA script that starts with a blank slate and does everything you'd otherwise do by hand. At that point you may as well write it in Python and generate the xlsx, ods or maybe just pdf for printing/visualization.

Most people, evidently yourself included, use spreadsheets like if they were a magical object that's guaranteed to read your mind and do exactly what you want. It's all too easy to fuck up if you're not careful, and unfortunately Excel by default doesn't come with policies that prevent you from fucking up. How one earth can anyone be sure that your projections aren't just line noise? Excel promotes development with no process, where there's no assurance about anything. That's one little reason why our financial industry is so fucked up. People trust Excel results, no questions asked.

Re:Live a day in my shoes

[sjbe]

You should know basic programming upon exiting high school.,

Programming has been a significant part of my job in years past. You have no idea what my background is.

So, what tool do you use to diff your spreadsheets? How do you ensure that there isn't a bug in a column of otherwise "identical" formulas?... blah blah

There are a multitude of ways to error check spreadsheets. RTFM. There also are plenty of tools to replicate formulas (or formatting or numbers). Furthermore you are inappropriately applying data modeling techniques that typically do not apply to what people need out of a spreadsheet. Spreadsheets aren't the right tool for every job but they are a great tool for many.

Spreadsheets provide a semblance of productivity and an illusion of efficiency.

Spoken well and truly like someone who doesn't do my job but thinks they know how to do it better than me. Seriously, you have no idea what you are talking about here. For what I use a spreadsheet for there simply is not a better tool available. If there were I would use that instead.

It's all too easy to fuck up if you're not careful, and unfortunately Excel by default doesn't come with policies that prevent you from fucking up.

The specific problems are different but there is NO programming tool that prevents mistakes.

That's one little reason why our financial industry is so fucked up. People trust Excel results, no questions asked.

No they don't and if you actually knew anything about the financial industry you would know that. You seem to have little idea what it is they are modeling and even less idea what level of precision is required for their models. People in finance use spreadsheets because they are GOOD ENOUGH for modeling the data they need. Many of them are perfectly capable of creating a database or using Matlab should the need arise but for their specific needs a spreadsheet is the best available tool for the job. Spreadsheets are easy to use, can provide a perfectly satisfactory tool for ad-hoc analysis (most of what they do), and they cost a reasonable amount of money.

Re:Live a day in my shoes

[tyrione]

No you should not know basic programming upon exiting high school. You should be required to have taken up to Calculus II and Linear Algebra with Physics, Chemistry and Biology if you plan to go to a university and be in the applied sciences. Learning programming is far less important.

Re:How is it?

[Patch86]

I use Calc all the time at home, and Excel all the time at work. Calc is great, but it's not as polished or feature-rich as Excel by a long shot.

Calc is good enough (more than good enough, in fact) for home use; it does everything I need it to, as quickly as I need it to, with features to spare. But Excel is still the better programme. If I had to do serious data crunching in Calc day in and day out, I'm sure it would drive me nuts.

Arguably, Excel is the only truly good programme in the MS Office suite (possible exception of Visio). I mean the rest (notably Word and PowerPoint) are fine and all, but there's just no clear benefit to them over the competition. There is literally not a single feature in Word that I can think of that is both useful to me and missing from LibreOffice; maybe there are word processor "power users" out there who might disagree with me, but for my usage I could easily live without it.

Re:Slashdort, listen up

[jfdavis668]

All GLORY TO THE HYPNOTOAD!

Appropriate tool use

[sjbe]

Thats not the issue. If your spreadsheet is SO larger that on a MODERN CPU, its slow ... you're doing it wrong.

It is a relatively trivial matter to make calculations on a dataset slow regardless of the tool used. I work with datasets and related calculations all the time that would make for slow calculations if you hand coded them in assembler. The mere fact that it is slow in a spreadsheet as well has nothing inherently to do with it being worked on in a spreadsheet. Now if the spreadsheet can't handle 65K rows by 65K columns then it shouldn't offer that size table as an option. But most can handle datasets that size and larger without too much trouble. For rapid data modeling and ad-hoc analysis a spreadsheet can be pretty hard to beat.

When people go wrong using spreadsheets it's usually one of a few ways. The one I see the most is when they take what should be a prototype analysis and turn it into a production tool. If you need to put a bunch of buttons and other interface tools on a spreadsheet THEN you are doing it wrong. The second is when they try to take analyzed data involving more than 3 dimensions. While it can be done it rarely is a good idea. Another I see is if they try to have more than one person working on the spreadsheet. If the dataset is truly huge or you require multi-user access or you need to interface with other applications then by all means use something other than a spreadsheet.

Re:Appropriate tool use

[K. S. Kyosuke]

Now if the spreadsheet can't handle 65K rows by 65K columns then it shouldn't offer that size table as an option.

It's not just a matter of mere number of cells. If the cell is "=A1*B2" and A1 and B2 are natural numbers, it probably doesn't pay off. But what if they are 2048x2048-sized matrices (for those spreadsheet applications that allow for that)? This way, you can either use more active cells, OR you can use more complicated calculations in the individual cells.

Re:Appropriate tool use

[McLoud]

When people go wrong using spreadsheets it's usually one of a few ways. The one I see the most is when they take what should be a prototype analysis and turn it into a production tool. If you need to put a bunch of buttons and other interface tools on a spreadsheet THEN you are doing it wrong. The second is when they try to take analyzed data involving more than 3 dimensions. While it can be done it rarely is a good idea.

Excel pivot`s is very good as a pivoting tool, and very few applications do it better, and I don't know of any that do so at excel's price range. Calc's version of it it's crap and kinda hard to setup.

Parallel computing

[sjbe]

Am I the only one that notices how crazy that sounds?

Why should it sound crazy? If you've got some parallel computations to make you'd be a fool not to use the GPU. There are many problems that could take advantage of the extra computing horsepower that are perfectly appropriate to do on a spreadsheet.

Re:Parallel computing

[0123456]

Why should it sound crazy? If you've got some parallel computations to make you'd be a fool not to use the GPU.

1. For simple tasks, parallel computing often ends up slower becuase the time taken to transfer data between processors is more than the time taken to do the calculations.
2. If you need teraflops of performance to process your spreadsheet, you're probably like my friend who used to write novels in Excel.

Re:Parallel computing

[TeXMaster]

Re. your point 1., OpenCL can run as well on CPUs (using all the cores and the vector instructions is 'free'), and of course the APUs, which can often access the same memory space as the CPU (especially with the upcoming hUMA designed by —guess who— AMD).

Re:Parallel computing

[Anonymous Coward]

Maybe there are some use cases between "simple tasks" and "need teraflops".

Re:How is it?

[Anonymous Coward]

My favorite excel function used to crash both LibreOffice and OpenOffince.The function in question is ={FREQUENCY(data ref,bins ref)}, but I do not know any other excel users that use array based functions. So for a long time, I naively thought that my use case was special, and that nobody but a few specialized workers needed Excel. I hold that pious but erroneous belief until I had to email a worksheet to a colleague running Linux. Sure, according to OO documentation that function is supported but it is was crashing his LO nonetheless...
LO should really put some work into that refactoring, being a sub par Excel is a far bigger adoption turn off than all the other LO/OO MsO compatibility problems.
Almost all executives do not care about a paragraph margin wrong by 2px but they all care about wrong numbers especially when those numbers are about money !!!

I like it

[Deathspawner]

It's times like this when I wish I actually had a need for such a thing. If LibreOffice ever allowed me to create prettier graphs like Word does, I'd consider moving on over. As much as Microsoft is hated on around here, Office is pretty damned polished (that isn't to say there are no problems... there are still many that drive me bonkers, but they are software features, not performance and the like).

Re:How is it?

[Anonymous Coward]

Access is the single most robust fast and dirty DB out there. Nothing else comes close to it on any platform.

The GPU is likely idle so why not use it?

[sjbe]

Indeed. You really shouldn't need to have to get a gaming GPU to run a spreadsheet. Hopefully

If you are doing trivial calculations then you are probably right. However many of us do more with spreadsheets than making grocery lists. There are quite a few problems that benefit from parallel processing. Since the GPU is probably sitting mostly idle if you have a spreadsheet up, why not do something useful with it?

Re:The GPU is likely idle so why not use it?

[hairyfeet]

Because its gonna use more power and generate more heat than the CPU alone would do? Unless you are using something LV like one of the AMD APUs but with those the memory bandwidth is gonna be a bottleneck and with the low end discretes frankly you aren't gonna get enough of a boost to be worth the trouble, probably nothing lower than an HD7750 (around $80) would be useful in this situation.

Re:How is it?

[Anonymous Coward]

I migrated to OO and then to LO and my financial spreadsheet worked great. Then came the updates and my file would no longer open. I bought MS Office through work for a very reduced price (we're talking $11) and the file opened just fine. Unfortunately, the stability issues I experienced were enough to turn me off those open-source options for at least a year or two more of development.

Re:How is it?

[MightyYar]

I love array functions. I mean, I hate them, but not as much as I hate writing VBA! I feel like MS could really make Excel more powerful if they improved the editing/editing/debugging of array functions. Maybe more people would use them if they were easier to construct.

Re:How is it?

[westlake]

I think my boss uses excel for simple formulas and for lists. I use Excel for anything not quite worthy of a Matlab script, so OpenOffice doesn't quite measure up for me but should work fine for my boss.

Your boss isn't interested in what you think.

What he needs is an office suite -- or integrated office system -- that can be deployed across the enterprise.

It doesn't matter if any single component is overkill for some so long as it works for all. Including temps, trainees, volunteers and so on.

Re:How is it?

[MightyYar]

My boss isn't stupid. She buys me a MATLAB and JMP license, and the design guys get CAD licenses. We are swimming in specialized software that she does not buy for herself. She only uses Excel because someone up the chain went with a site license for MS Office. If they hadn't made that decision, she'd use whatever made the next-best list maker (I believe she used Quatro Pro prior to our site license). We also use the abysmal Sharepoint system as a glorified shared drive, just because it is there.

MVC

[Bill_the_Engineer]

Just answering a question in your comment.

The basic principle was separation of data, views, and formulas (was this pre-figuring MVC?),

MVC was introduced in 1978's as part of the work by Xerox PARC.

What about WRITER?

[Anonymous Coward]

I saw this headline and the first thing in my mind was "what about libre office writer"? I have just finish my diploma thesis and while it was not that big(less than 100 pages), the program was slow. By slow I mean when I was scrolling from one page to another writer froze for 5-7 seconds. It's true that I had a lot of pictures in the document but that should be a non issue on an 3770k cpu. Seriously...what's up with this slowness. (If there is a way to fix this please leave me a comment to tell me how)

What about writer ?

[holo4ever]

I saw this headline and the first thing in my mind was "what about libre office writer"? I have just finish my diploma thesis and while it was not that big(less than 100 pages), the program was slow. By slow I mean when I was scrolling from one page to another writer froze for 5-7 seconds. It's true that I had a lot of pictures in the document but that should be a non issue on an 3770k cpu. Seriously...what's up with this slowness. (If there is a way to fix this please leave me a comment to tell me how Wanted to repost this under myself and not AC. I noticed my posts don't show up.... :/

Doesn't libreoffice have bigger problems than that

[Mirar]

So...
1. add 64-bit support, so you can have large documents? (Windows.) https://bugs.freedesktop.org/show_bug.cgi?id=61683
2. fix the actual code, so it doesn't take forever to do almost nothing?

What does the GPU have to do with that? :p

Nice that people are working on it though.

I haven't really noticed

[Anonymous Coward]

I haven't really noticed that Calc is slow, even though I used it to calculate how much money I needed to repay the bank after getting my CS (I calculated the interest by month, how much each payment was, how much went to the bank and how little went to paying down the principle. You can run simulations (like if the bank tells you to open a savings account while paying down a loan, exactly how stupid that advice is, how much more you will wind up paying them, how it only works in your favor if you get a better rate of interest than what you are paying them, etc.). I've also used these spreadsheets over the past several years to create forms that look a lot like the tax forms I get from the government, except with smarts put in (by me) to do calculations. I file, and have not been audited in years. I agree that creating a database to do a 'one shot' thing like taxes is a big waste of time, and I hadn't noticed how long calc takes (my biggest spreadsheets are only 5 or 6 pages, and never more than 150 lines down by about 20 rows across). I cross reference sheets quite a bit (an entry on one sheet links to entries on other sheets), and use calculations and if/then/else logic and that's about as complex as it gets. If they want to make it faster, go hard.

Re:Right (Re:If you need it you are doing it wrong

[tyrione]

Actually, the UI for Lotus Improv was quite nice and won some awards.

Its (spiritual) successor, Quantrix Financial Modeler seems to be selling well enough, even w/ a $1,495 price point.

I wish that Flexisheet (an opensource take on this sort of thing) would get more traction.

Correct and Improv was written specifically for NeXTSTEP which made it possible to do what it did. Improv fell apart and couldn't replicate the same MVC frameworks of ObjC/NeXTSTEP on Windows. Quantrix Financial Modeler was purchased and was also originally on NeXTSTEP. I used both at NeXT.

Unit Tests

[caspy7]

...unit tests are being added for the first time...

Wha?
How can a project so large (and so old) not have unit tests on its spreadsheet application?
This seems like it would be a basic necessity from the beginning.

This is great news

[FithisUX]

It makes sense. FPU is dead long live the multi-core FPUs exported by OpenCL

Don't trust important calculations to toys

[dakra137]

Are the results correct? Well I've double checked the formulae.

So what, if your computer doesn't use Error Correcting Code (ECC) Memory or at least parity memory, soft or hard errors may not be noticed.

If you must use toy class computers for important work:
Verify the data and formulae and calculate file hashes on a business class computer,

for (i=1; i < 3; i++) {transfer to toy[i]; verify the hashes; calculate;}
Gather the results; compare ;
if (same) {hurray;} else {try again;}