Well, maybe the mistake is picking micro brews that are too dark. Their is a lot of space between the water that is pabst, and the overtoasted crap on the other side of the spectrum.
I find that if I can't see my hand through the pint, the beer is hiding something. There is a reason a lot of home brewers do stout, and it has to do with their inability to make a good IPA or amber ale. Those lighter beers are very susceptible to off tastes if everything isn't super clean, and the trub isn't removed fairly quickly.
BTW: Finally, any beer that is lighter than a light amber color is probably heavily cut with rice or corn... In the case of bud light/etc there is so little barley or hops its hard to call it "beer" with a straight face.
I doubt it, when your apple and your buying the entire supply of something you get quite a nice price break.
Anyway, I would be curious to see them do some price comparisons with those $20 DVD players at walmart, or the $99 kids tablet/netbooks at toys R US, or the $200 netbooks at frys. Or for that matter the $50 capacitive screen tablets that are everywhere. Basically devices that don't have some big company subsidizing them to gain further sales.
Trying to legislate them away is the kind of stupidity I expect from our government.
Its as dumb as what I did outside. I have a couple of exterior house lights with security settings that turn them on at dusk and then I can set how long they stay on for at 100% or dim to 50%, as well as how long they stay on if the motion detector triggers. Well these lamps 100% don't work with anything other than 100W incandescent bulbs. The dim-able CFL's and LED's both choke or blow up in and hour or two. So, my solution is I swapped in a 23W CFL, and wired it to bypass the controller. The plan was to remember to turn it on/off, but I'm not that kind of person, so it just runs 100% of the time.
I'm sure the energy savings in this case doesn't exist. Couple hours a night at 100W vs 24 hours a day at 23W... Good job there...
Or they are just to stupid to remember the windows logo program, which basically are/were a bunch of stickers that said "designed for windows 98" plastered all over hardware. Heck the old HP laptops I have linux installed on have stickers on them that say "Designed for windows 2000" which basically means that the devices meet the logo requirements for windows 2000, which was something like a 200Mhz CPU,16M or RAM, and color coded ports on the back.
When they decided to call the 80586 a Pentium to keep cyrix, harris, amd, idt, etc from naming their products the same.
But, now days for software it seems like if you hire a marketdroid they have to justify their existence by renaming your product/company/etc every year or two.
Your response to me that the articles are written by grad students and mathemeticians (not all mathemeticians are jerks, btw) for grad students and mathemeticians reinforces the fact that it certainly seems like a giant circle jerk.
I wrote a couple of the original pages on wikipedia dealing with some comp-sci type topics not usually taught in a 4 year program (back in ~2000). I thought they were fairly clear and understandable, complete with short pieces of pseudo code, and algorithmic explanations. I gave up on those pages because of the idiots editing them. I couple years ago I looked at them, and frankly was shocked, its like some grad students have been trying to out do each other on writing the most esoteric mathematical description of the problem/solution. In fact the Wikipedia pages probably should just be wiped and replaced with references to the original authors (of the algorithms/mathematics) works, because they are a far more accessible resource.
Basically, I would call the Wikipedia pages a complete failure on any metric other than a competition to obfuscate with math something that is fairly accessible.
because the locals, who refused to fund it with taxes
I was there when those roads were being planned, the locals were very angry about them being toll roads. The state has consistently refused to invest in Austin area roads (compared with the rest of the state). Perry and the Legislature rammed it through anyway. Partially because it was their buddies getting the contracts to operate them. Its operated differently than every other toll road in the US, where the tolls go into the highway budget, instead those roads are a "private public partnership" which is code for all the profits going to big corps. http://www.humanevents.com/2007/03/12/texas-toll-road-plan-stirs-grassroots-protest/
In fact as another poster pointed out they were intending to add tolls to existing tax payer funded roads. Plus, the amount they can charge is based on some formulas which have a speed limit component.
Basically, the speed limit is being bumped so Perry's buddies can make more money.
6 W? That's a lot, just to keep a few DIMMs and some helper electronics going...
Yes, but I think most of it is the power supplies being 80% inefficient at that low of current draw. If someone were to hard power the monitors off when the PC goes to sleep it would probably save a watt or two there too. Its actually more than my (ARM based) NAS running full out, but I'm pretty happy though.
I don't think windows saves/restores the entire RAM range on hibernate/resume. AFAIK, only the kernel/non pagable and dirty pages are flushed, the rest is simply marked paged out. Then when you resume it reverse the whole thing and starts to fault in pages.
So the average user probably isn't initially moving more than a hundred or two megs of stuff in a multi gigabyte system before the system starts to respond. Especially if most of the ram was disk cache (or other garbage that won't be saved/restored).
That is why the system seems to beat the disk up for a few seconds/minutes after a resume and tends to be a little sluggish at first.
I've been using S3 standby on a desktop with a UPS for the last 9 years or so with windows XP.
Fantastic, basically i never reboot XP, walk away it goes to sleep after 15 minutes, and wakes on a key press in under 1 second.
Power draw in S3 is ~6W including monitors after I purchased a 80+ power supply I found on some website talking abut standby power usage (most of them are only 80+ efficient in the intended power range which is generally > 50 watts).
Seriously? 80 instructions per clock cycle, sustained? Bullshit.
Thats caused by people multiplying the single core numbers by the core/thread count. You can easily buy an 80 core x86's (HP DL980G7). Which leaves the IPC at only 1 (using your number of 80 IPC), which is probably actually really low for that machine considering the cores are super-scalar, and its hyperthreaded.
That is in fact what IBM does with their zseries too, the single core numbers are only like 1.3k MIPS, times 64 or whatnot for the bigger configurations.
Generally, the larger the memory system the longer the memory latency. Hence the need for larger caches. The one thing the z does bring to the table is extremely large caches. Which helps a lot if you are sitting in that range between fitting in cache or not. The problem is (at least with the previous generation CPU) is that the memory latency is not as good as you assume. You can't throw money at the latency problem, like you can with the bandwidth or capacity ones.
From what I can tell from my z114 the memory latency isn't anything to write home about. Its ok, but its is _NOT_ better than my mid range DL380's. So your example sort of sucks. In fact with x86 servers the fewer sockets you have generally translates to better latency because the data has to be transferred over the QPI or HT links. So if you app is bound by memory latency you actually want a smaller machine (unless you get small enough your hitting disk...).
Perhaps there are no z/OS drivers for them, but Linux drivers might exist.
The only PCIe adapters you can plug in are ones sold by IBM for the z. Getting them working under zos is pretty much a non starter, zos is _STUPID_ it can't even talk to normal FC attached disks even though the adapter can do it under linux.
If you want to have a lot of fun hacking a z PCIe layer into linux you might get it working. That is after you hack up zvm to allow passthough, or the LPAR code... Linux is only sort of running on the hardware, think vmware without pci passthrough. Basically, you need to be able to modify the equivalent of vmware before linux even gets a view of the device.
This is what lspci says when I run it under sles on a z114 stuffed with pcie adapters.
pcilib: Cannot open/proc/bus/pci lspci: Cannot find any working access method.
This is/sys/bus drwxr-xr-x 4 root root 0 Jun 12 15:10 ccw/ drwxr-xr-x 4 root root 0 Jun 12 15:10 ccwgroup/ drwxr-xr-x 4 root root 0 Jun 12 15:10 css/ drwxr-xr-x 4 root root 0 Jun 12 15:10 event_source/ drwxr-xr-x 4 root root 0 Jun 12 15:10 platform/ drwxr-xr-x 4 root root 0 Jun 12 15:10 scsi/ drwxr-xr-x 4 root root 0 Jun 12 15:10 virtio/
So what benchmark did you run to determine that a Pentium P90 had the same performance as a base z114?
A bunch of in house stuff, as well as some standalone open source benchmarks I could get my hands on, and would compile under OMVS. I use the P90 example because that is what nbench scored the single core performance as (single threaded, fairly small working set), and the variation in its results were fairly consistent with a number of other benchmarks I was running. Obviously changing the weighting could shift the benchmarks significantly in either direction as there was a couple x variation in the results. The IO benchmarks were a different beast, but in many cases there was sufficient CPU load that they were CPU bound, enough that the aggregate raw IO and throughput rates were what I would call appropriate in the 1990s. If you want to donate a SPEC license....
I also tested basically the same things under a SLES install running on some IFL's to get a feeling for the raw system performance. It was significantly better (duh!), but not enough that I would consider it a valid machine to host actual workloads even without the price tag. If a fully licensed 3k MIPS machine were priced in line with a lower midrange x86 then it might be a contender, but I can buy a whole rack of high end x86s, and a decent disk subsystem for the prices IBM wants for even a midrange z114. My seat of the pants estimates put the z at about 10-20x the cost for similar performance.
Basically, if your a big bank who cares what it costs. But for most companies outside of the fortune 100 throwing that kind of money away on a fairly small problem is probably hard to justify. You want to see what real high transaction rate systems look like today, just look at google, facebook and amazon.
You can buy 16Gbit FC from IBM for x series. The z only has 8G. Ethernet is 10G on OSA, 40G Ethernet is available.
Basically, the open systems peripheral market is where all the cutting edge stuff is. No one who has a great idea for how to make a faster/better IO adapter limits themselves to selling 100 of them into the mainframe space. Especially when they can get purchased for billions of dollars by the likes of Dell, EMC and HP.
When it comes to SSD's I was talking solid state, which can be attached via FC on the z, but the performance is nowhere near what can be achieved with something like the fusionIO PCIe based adapters. FICON really limits the per pipe IOP rates, so you end up throwing dozens and dozens of ports into the mix to get the IO rates up, especially if there is any IO latency.
Then you buy a high end Fiber channel or infiniband disk array for your x86. Or for that matter you go cheap and buy a fusionIO, and beat an 8 figure mainframe. IO on a modern x86 is usually limited by the peripheral, not by the bus or memory bandwidth. Its fairly easy to drive 10's of GB (bytes!)/sec of IO bandwidth on an x86. The problem is finding a network link to support it, or a disk farm to source it from.
There is no guarantee that Dhrystone "MIPS" numbers correspond at all to IBM "MIPS" numbers.
Correct, a MI (million instructions) on the z, isn't equal to a MI on any other platform. That said IBM rates their machines in "MIPS", for comparison purposes. A P90 is rated at roughly 70 Dhrystone MIPS per (http://www.alternatewars.com/BBOW/Computing/Computing_Power.htm). When I stated they there are roughly the same performance I meant it. So, you can use that as a conversion factor. Given the inaccuracies of the benchmarks/etc, you could just multiply a mainframe MIP by 3 to get a rough approximation to a x86 dhrystone MI. Of course any given benchmark is going to be a little higher or lower than this depending on what its doing.
here are the competitive systems you seem to think are possible? Quit your job and go make some. Just don't collect unemployment when you fail.
BTW: My day job in the early 90's was working for a very successful company selling a niche application to county/city governments. In nearly every case, we were replacing mainframes. Good luck finding a mainframe in that particular space. We had a couple competitors doing exactly the same thing. By the late 1990's finding a mainframe in that space was hard. They are gone, replaced by unix servers. Outside of some really rich companies in the banking/insurance industries its difficult to find a mainframe today. For every company you name running a mainframe in their core business I can probably name two running an AS400, or an AIX machine. The companies still running mainframes aren't doing it because zos is better, or the machine is more powerful, or it can handle more TPM.
I'm quite sure that for the applications people actually use mainframes for, you're utterly wrong. Not only do they scale massively higher in terms of throughput, they also manage to do it with obscene uptimes (measured in years) and reliability nothing can compare to.
No, this is where you are wrong, open system are far more advanced in both processing and IO capabilities. In fact with the latest x86's the RAS features are equivalent or better. If you had actually maintained a mainframe for any length of time you wouldn't be spewing this. They get IPL'ed (rebooted) regularly just like any other compute device in the data center. Sure you may only be IPL'ing an LPAR, but that isn't any different than what I do with vmware when I reboot or migrate an image. In fact vmware vmotion is actually is more robust than anything i've seen on z.
Modern SSD, FC, ethernet and inifiniband connections on x86 are light years beyond the mainframe. The idea of off loading IO operations to channel processors, has been on x86 hardware for years now. Modern x86 disk IO is little more than queue the request with a DMA pointer, and wait for the hardware to notify you its in memory. Any given x86 server probably has a half dozen embedded ARM's or whatnot sitting the the ASICs from broadcom/qlogic/etc. Companies like FusionIO, make PCIe SSD's that by themselves allow entry level x86 servers to compete on IOP benchmarks with 8 figure mainframes.
Besides there is more than x86, IBM's own POWER is better, at significantly less cost.
I've worked on projects trying to do exactly this. And I've seen a couple of them fail.
Trying to map out all of the use cases for software which is mission critical and has been around since the 60's can actually prove to be exceedingly challenging if not impossible.
This is actually, where your sorta right. Like any other software project migrations of proprietary apps can be extremely painful. That said, lots of times, the migrations are failing not because of the migration itself, but that people are trying to fix all the problems with the mainframe software at the same time. Rarely do this failures have anything to do with the technology of the mainframe vs anything else. Instead they fail for all the same reasons any other large software project fails. The return on doing it though is almost easy to justify, which is why there are hundreds of companies doing exactly this. One of my previous jobs was similar, we would walk in with a pair of x86 servers and replace a 7-8 figure mainframe. Our whole solution cost less than a month of maintenance, not counting the army of support personnel, maintenance, licensing, etc.
Those are emulated cores under hercules. If you were running oracle you would run them natively on the cores, and in that case its closer to 1:1 with the mainframe.
5.5Ghz probably makes it about as fast as a 2 year old intel machine. I should know, I have a z114 (previous generation at 3.8Ghz) that i've done extensive benchmarks on. The fact that IBM refuses to publish standard benchmark numbers (specCPU, specVM, etc) should be sufficient proof that they are not pretty.
I can say that the people buying these things are pretty much smoking some fine IBM drugs. Sure, they are actually fairly competitive (but still not class leading) on the high end, but on the low end, which starts at ~200k, after disks and licenses, for 26 MIPS are abysmal. At that price/performance hercules on a midrange desktop PC doing software emulation (and its not even JIT'ed) runs somewhere between 5-15x as fast.
A 26 MIP mainframe is roughly equal to a Pentium 90. A full blown 3.8 Ghz z114 is roughly equal to a 5 year old x86 server.
Worse yet, is FICON, which generally is just a giant layer of inefficiency sitting in front of standard SCSI/SAS disks. So, the IO numbers are pretty abysmal too.
Basically, you have to spend >$400k before the mainframe catches up to what you can do on your desktop with a free emulator.
If your running linux on z, then your really deluded. In fact, your probably better off taking the HMCs, SEs, and CUs that it comes with and running linux on them directly. The only minor saving grace is that IBM doesn't rape people for unlocked processors to run linux (IFL's).
Further, IBM's claims of easier manageability are a joke. I can install ESXi and a half dozen linux machines, in the time it takes an expert system programmer to setup the HCD, install z/vm, and start configuring a linux machine. Oh, and I can migrate the image with a couple mouse clicks. Plus, I don't have to manage my data stores as a bunch of tiny disk images because zOS still prefers to deal with mod3 (~3GB) and mod9 (~9GB) disk partitions. I literally have a few hundred partitions on a machine with just a couple TB of storage. If you think managing a few dozen vmware disks is a problem, multiply it by 3-8x on z to run linux.
Frankly, if you have cobol, JCL, whatever running on these things and your not desperately trying to migrate to another platform, then your must either be extremely rich, or really stupid. The maintenance costs alone over ten years is going to save 7 figure sums, which should more more than enough to hire a couple programmers and a system administrator to port and maintain the apps on a machine that costs $20k every 5 years.
The separate processes for each tab is EXACTLY what makes Chrome superior.
Actually, its one of the reasons I don't run it.
Call me stupid, but I have this nasty habit of overusing "open in new tab" and clicking a bunch of links I never end up reading. Then leaving my browser windows open because "damit there is still some cool things I need to read in those tabs"
The end result is that after a few days my browser is basically consuming all the memory/CPU that is available. With a 32-bit firefox that is just a couple of gigs and a single CPU. The remainder of the system remains responsive. With chrome, it brings the whole OS to its knees, swapping and thrashing. And, yah my desktop machines all have between 8 and 24GB of RAM, but they also are running other apps that consume a fair amount of it (vmware for example).
need massive amounts of HDD I/O at all? And how did this even get marked "insightful"?
Its called paging, and it happens when you run out of ram, this could be because you have other apps open also consuming gigs of RAM. Its also particularly problematic because chrome is susceptible to javascript memory leaks like FF is. Actually, it might be worse in a couple cases in chrome (direct experience trying to fix a memory leak in one of our web apps). There is nothing like showing up in the morning at work, and your machine is crawling because your browser as gone and consumed all available RAM due to a couple leaky ajax handlers.
Your using chrome as an example? Sure its fast, but its also 6M LOC. The main reason apple chose KHTML/KJS was that it was easy to maintain at something like 150K LOC.
The problem is who are they going to replace him with, look at what has being going on at HP for a decade. That is another blown tech story, and fundamentally reflects what happens when your management is out of touch in the tech industry. You might be able to sell shitty cars for decades before anyone wakes up and discovers your product and company is total shit (GM!), but in the tech industry it happens much faster.
Slashdot Mirror
Well, maybe the mistake is picking micro brews that are too dark. Their is a lot of space between the water that is pabst, and the overtoasted crap on the other side of the spectrum.
I find that if I can't see my hand through the pint, the beer is hiding something. There is a reason a lot of home brewers do stout, and it has to do with their inability to make a good IPA or amber ale. Those lighter beers are very susceptible to off tastes if everything isn't super clean, and the trub isn't removed fairly quickly.
BTW: Finally, any beer that is lighter than a light amber color is probably heavily cut with rice or corn... In the case of bud light/etc there is so little barley or hops its hard to call it "beer" with a straight face.
I doubt it, when your apple and your buying the entire supply of something you get quite a nice price break.
Anyway, I would be curious to see them do some price comparisons with those $20 DVD players at walmart, or the $99 kids tablet/netbooks at toys R US, or the $200 netbooks at frys. Or for that matter the $50 capacitive screen tablets that are everywhere. Basically devices that don't have some big company subsidizing them to gain further sales.
Trying to legislate them away is the kind of stupidity I expect from our government.
Its as dumb as what I did outside. I have a couple of exterior house lights with security settings that turn them on at dusk and then I can set how long they stay on for at 100% or dim to 50%, as well as how long they stay on if the motion detector triggers. Well these lamps 100% don't work with anything other than 100W incandescent bulbs. The dim-able CFL's and LED's both choke or blow up in and hour or two. So, my solution is I swapped in a 23W CFL, and wired it to bypass the controller. The plan was to remember to turn it on/off, but I'm not that kind of person, so it just runs 100% of the time.
I'm sure the energy savings in this case doesn't exist. Couple hours a night at 100W vs 24 hours a day at 23W... Good job there...
Or they are just to stupid to remember the windows logo program, which basically are/were a bunch of stickers that said "designed for windows 98" plastered all over hardware. Heck the old HP laptops I have linux installed on have stickers on them that say "Designed for windows 2000" which basically means that the devices meet the logo requirements for windows 2000, which was something like a 200Mhz CPU,16M or RAM, and color coded ports on the back.
Its probably a misunderstanding...
When they decided to call the 80586 a Pentium to keep cyrix, harris, amd, idt, etc from naming their products the same.
But, now days for software it seems like if you hire a marketdroid they have to justify their existence by renaming your product/company/etc every year or two.
Your response to me that the articles are written by grad students and mathemeticians (not all mathemeticians are jerks, btw) for grad students and mathemeticians reinforces the fact that it certainly seems like a giant circle jerk.
I wrote a couple of the original pages on wikipedia dealing with some comp-sci type topics not usually taught in a 4 year program (back in ~2000). I thought they were fairly clear and understandable, complete with short pieces of pseudo code, and algorithmic explanations. I gave up on those pages because of the idiots editing them. I couple years ago I looked at them, and frankly was shocked, its like some grad students have been trying to out do each other on writing the most esoteric mathematical description of the problem/solution. In fact the Wikipedia pages probably should just be wiped and replaced with references to the original authors (of the algorithms/mathematics) works, because they are a far more accessible resource.
Basically, I would call the Wikipedia pages a complete failure on any metric other than a competition to obfuscate with math something that is fairly accessible.
because the locals, who refused to fund it with taxes
I was there when those roads were being planned, the locals were very angry about them being toll roads. The state has consistently refused to invest in Austin area roads (compared with the rest of the state). Perry and the Legislature rammed it through anyway. Partially because it was their buddies getting the contracts to operate them. Its operated differently than every other toll road in the US, where the tolls go into the highway budget, instead those roads are a "private public partnership" which is code for all the profits going to big corps. http://www.humanevents.com/2007/03/12/texas-toll-road-plan-stirs-grassroots-protest/
In fact as another poster pointed out they were intending to add tolls to existing tax payer funded roads. Plus, the amount they can charge is based on some formulas which have a speed limit component.
Basically, the speed limit is being bumped so Perry's buddies can make more money.
6 W? That's a lot, just to keep a few DIMMs and some helper electronics going...
Yes, but I think most of it is the power supplies being 80% inefficient at that low of current draw. If someone were to hard power the monitors off when the PC goes to sleep it would probably save a watt or two there too. Its actually more than my (ARM based) NAS running full out, but I'm pretty happy though.
I don't think windows saves/restores the entire RAM range on hibernate/resume. AFAIK, only the kernel/non pagable and dirty pages are flushed, the rest is simply marked paged out. Then when you resume it reverse the whole thing and starts to fault in pages.
So the average user probably isn't initially moving more than a hundred or two megs of stuff in a multi gigabyte system before the system starts to respond. Especially if most of the ram was disk cache (or other garbage that won't be saved/restored).
That is why the system seems to beat the disk up for a few seconds/minutes after a resume and tends to be a little sluggish at first.
I've been using S3 standby on a desktop with a UPS for the last 9 years or so with windows XP.
Fantastic, basically i never reboot XP, walk away it goes to sleep after 15 minutes, and wakes on a key press in under 1 second.
Power draw in S3 is ~6W including monitors after I purchased a 80+ power supply I found on some website talking abut standby power usage (most of them are only 80+ efficient in the intended power range which is generally > 50 watts).
Seriously? 80 instructions per clock cycle, sustained? Bullshit.
Thats caused by people multiplying the single core numbers by the core/thread count. You can easily buy an 80 core x86's (HP DL980G7). Which leaves the IPC at only 1 (using your number of 80 IPC), which is probably actually really low for that machine considering the cores are super-scalar, and its hyperthreaded.
That is in fact what IBM does with their zseries too, the single core numbers are only like 1.3k MIPS, times 64 or whatnot for the bigger configurations.
Generally, the larger the memory system the longer the memory latency. Hence the need for larger caches. The one thing the z does bring to the table is extremely large caches. Which helps a lot if you are sitting in that range between fitting in cache or not. The problem is (at least with the previous generation CPU) is that the memory latency is not as good as you assume. You can't throw money at the latency problem, like you can with the bandwidth or capacity ones.
From what I can tell from my z114 the memory latency isn't anything to write home about. Its ok, but its is _NOT_ better than my mid range DL380's. So your example sort of sucks. In fact with x86 servers the fewer sockets you have generally translates to better latency because the data has to be transferred over the QPI or HT links. So if you app is bound by memory latency you actually want a smaller machine (unless you get small enough your hitting disk...).
Perhaps there are no z/OS drivers for them, but Linux drivers might exist.
The only PCIe adapters you can plug in are ones sold by IBM for the z. Getting them working under zos is pretty much a non starter, zos is _STUPID_ it can't even talk to normal FC attached disks even though the adapter can do it under linux.
If you want to have a lot of fun hacking a z PCIe layer into linux you might get it working. That is after you hack up zvm to allow passthough, or the LPAR code... Linux is only sort of running on the hardware, think vmware without pci passthrough. Basically, you need to be able to modify the equivalent of vmware before linux even gets a view of the device.
This is what lspci says when I run it under sles on a z114 stuffed with pcie adapters.
pcilib: Cannot open /proc/bus/pci
lspci: Cannot find any working access method.
This is /sys/bus
drwxr-xr-x 4 root root 0 Jun 12 15:10 ccw/
drwxr-xr-x 4 root root 0 Jun 12 15:10 ccwgroup/
drwxr-xr-x 4 root root 0 Jun 12 15:10 css/
drwxr-xr-x 4 root root 0 Jun 12 15:10 event_source/
drwxr-xr-x 4 root root 0 Jun 12 15:10 platform/
drwxr-xr-x 4 root root 0 Jun 12 15:10 scsi/
drwxr-xr-x 4 root root 0 Jun 12 15:10 virtio/
Never underestimate how stupid this machine is.
So what benchmark did you run to determine that a Pentium P90 had the same performance as a base z114?
A bunch of in house stuff, as well as some standalone open source benchmarks I could get my hands on, and would compile under OMVS. I use the P90 example because that is what nbench scored the single core performance as (single threaded, fairly small working set), and the variation in its results were fairly consistent with a number of other benchmarks I was running. Obviously changing the weighting could shift the benchmarks significantly in either direction as there was a couple x variation in the results. The IO benchmarks were a different beast, but in many cases there was sufficient CPU load that they were CPU bound, enough that the aggregate raw IO and throughput rates were what I would call appropriate in the 1990s. If you want to donate a SPEC license....
I also tested basically the same things under a SLES install running on some IFL's to get a feeling for the raw system performance. It was significantly better (duh!), but not enough that I would consider it a valid machine to host actual workloads even without the price tag. If a fully licensed 3k MIPS machine were priced in line with a lower midrange x86 then it might be a contender, but I can buy a whole rack of high end x86s, and a decent disk subsystem for the prices IBM wants for even a midrange z114. My seat of the pants estimates put the z at about 10-20x the cost for similar performance.
Basically, if your a big bank who cares what it costs. But for most companies outside of the fortune 100 throwing that kind of money away on a fairly small problem is probably hard to justify. You want to see what real high transaction rate systems look like today, just look at google, facebook and amazon.
Yes, they tend to lag about a generation.
You can buy 16Gbit FC from IBM for x series. The z only has 8G. Ethernet is 10G on OSA, 40G Ethernet is available.
Basically, the open systems peripheral market is where all the cutting edge stuff is. No one who has a great idea for how to make a faster/better IO adapter limits themselves to selling 100 of them into the mainframe space. Especially when they can get purchased for billions of dollars by the likes of Dell, EMC and HP.
When it comes to SSD's I was talking solid state, which can be attached via FC on the z, but the performance is nowhere near what can be achieved with something like the fusionIO PCIe based adapters. FICON really limits the per pipe IOP rates, so you end up throwing dozens and dozens of ports into the mix to get the IO rates up, especially if there is any IO latency.
But what if you need I/O?
Then you buy a high end Fiber channel or infiniband disk array for your x86. Or for that matter you go cheap and buy a fusionIO, and beat an 8 figure mainframe. IO on a modern x86 is usually limited by the peripheral, not by the bus or memory bandwidth. Its fairly easy to drive 10's of GB (bytes!)/sec of IO bandwidth on an x86. The problem is finding a network link to support it, or a disk farm to source it from.
There is no guarantee that Dhrystone "MIPS" numbers correspond at all to IBM "MIPS" numbers.
Correct, a MI (million instructions) on the z, isn't equal to a MI on any other platform. That said IBM rates their machines in "MIPS", for comparison purposes. A P90 is rated at roughly 70 Dhrystone MIPS per (http://www.alternatewars.com/BBOW/Computing/Computing_Power.htm). When I stated they there are roughly the same performance I meant it. So, you can use that as a conversion factor. Given the inaccuracies of the benchmarks/etc, you could just multiply a mainframe MIP by 3 to get a rough approximation to a x86 dhrystone MI. Of course any given benchmark is going to be a little higher or lower than this depending on what its doing.
here are the competitive systems you seem to think are possible? Quit your job and go make some. Just don't collect unemployment when you fail.
BTW: My day job in the early 90's was working for a very successful company selling a niche application to county/city governments. In nearly every case, we were replacing mainframes. Good luck finding a mainframe in that particular space. We had a couple competitors doing exactly the same thing. By the late 1990's finding a mainframe in that space was hard. They are gone, replaced by unix servers. Outside of some really rich companies in the banking/insurance industries its difficult to find a mainframe today. For every company you name running a mainframe in their core business I can probably name two running an AS400, or an AIX machine. The companies still running mainframes aren't doing it because zos is better, or the machine is more powerful, or it can handle more TPM.
I'm quite sure that for the applications people actually use mainframes for, you're utterly wrong. Not only do they scale massively higher in terms of throughput, they also manage to do it with obscene uptimes (measured in years) and reliability nothing can compare to.
No, this is where you are wrong, open system are far more advanced in both processing and IO capabilities. In fact with the latest x86's the RAS features are equivalent or better. If you had actually maintained a mainframe for any length of time you wouldn't be spewing this. They get IPL'ed (rebooted) regularly just like any other compute device in the data center. Sure you may only be IPL'ing an LPAR, but that isn't any different than what I do with vmware when I reboot or migrate an image. In fact vmware vmotion is actually is more robust than anything i've seen on z.
Modern SSD, FC, ethernet and inifiniband connections on x86 are light years beyond the mainframe. The idea of off loading IO operations to channel processors, has been on x86 hardware for years now. Modern x86 disk IO is little more than queue the request with a DMA pointer, and wait for the hardware to notify you its in memory. Any given x86 server probably has a half dozen embedded ARM's or whatnot sitting the the ASICs from broadcom/qlogic/etc. Companies like FusionIO, make PCIe SSD's that by themselves allow entry level x86 servers to compete on IOP benchmarks with 8 figure mainframes.
Besides there is more than x86, IBM's own POWER is better, at significantly less cost.
I've worked on projects trying to do exactly this. And I've seen a couple of them fail.
Trying to map out all of the use cases for software which is mission critical and has been around since the 60's can actually prove to be exceedingly challenging if not impossible.
This is actually, where your sorta right. Like any other software project migrations of proprietary apps can be extremely painful. That said, lots of times, the migrations are failing not because of the migration itself, but that people are trying to fix all the problems with the mainframe software at the same time. Rarely do this failures have anything to do with the technology of the mainframe vs anything else. Instead they fail for all the same reasons any other large software project fails. The return on doing it though is almost easy to justify, which is why there are hundreds of companies doing exactly this. One of my previous jobs was similar, we would walk in with a pair of x86 servers and replace a 7-8 figure mainframe. Our whole solution cost less than a month of maintenance, not counting the army of support personnel, maintenance, licensing, etc.
Those are emulated cores under hercules. If you were running oracle you would run them natively on the cores, and in that case its closer to 1:1 with the mainframe.
5.5Ghz probably makes it about as fast as a 2 year old intel machine. I should know, I have a z114 (previous generation at 3.8Ghz) that i've done extensive benchmarks on. The fact that IBM refuses to publish standard benchmark numbers (specCPU, specVM, etc) should be sufficient proof that they are not pretty.
I can say that the people buying these things are pretty much smoking some fine IBM drugs. Sure, they are actually fairly competitive (but still not class leading) on the high end, but on the low end, which starts at ~200k, after disks and licenses, for 26 MIPS are abysmal. At that price/performance hercules on a midrange desktop PC doing software emulation (and its not even JIT'ed) runs somewhere between 5-15x as fast.
A 26 MIP mainframe is roughly equal to a Pentium 90. A full blown 3.8 Ghz z114 is roughly equal to a 5 year old x86 server.
Worse yet, is FICON, which generally is just a giant layer of inefficiency sitting in front of standard SCSI/SAS disks. So, the IO numbers are pretty abysmal too.
Basically, you have to spend >$400k before the mainframe catches up to what you can do on your desktop with a free emulator.
If your running linux on z, then your really deluded. In fact, your probably better off taking the HMCs, SEs, and CUs that it comes with and running linux on them directly. The only minor saving grace is that IBM doesn't rape people for unlocked processors to run linux (IFL's).
Further, IBM's claims of easier manageability are a joke. I can install ESXi and a half dozen linux machines, in the time it takes an expert system programmer to setup the HCD, install z/vm, and start configuring a linux machine. Oh, and I can migrate the image with a couple mouse clicks. Plus, I don't have to manage my data stores as a bunch of tiny disk images because zOS still prefers to deal with mod3 (~3GB) and mod9 (~9GB) disk partitions. I literally have a few hundred partitions on a machine with just a couple TB of storage. If you think managing a few dozen vmware disks is a problem, multiply it by 3-8x on z to run linux.
Frankly, if you have cobol, JCL, whatever running on these things and your not desperately trying to migrate to another platform, then your must either be extremely rich, or really stupid. The maintenance costs alone over ten years is going to save 7 figure sums, which should more more than enough to hire a couple programmers and a system administrator to port and maintain the apps on a machine that costs $20k every 5 years.
The separate processes for each tab is EXACTLY what makes Chrome superior.
Actually, its one of the reasons I don't run it.
Call me stupid, but I have this nasty habit of overusing "open in new tab" and clicking a bunch of links I never end up reading. Then leaving my browser windows open because "damit there is still some cool things I need to read in those tabs"
The end result is that after a few days my browser is basically consuming all the memory/CPU that is available. With a 32-bit firefox that is just a couple of gigs and a single CPU. The remainder of the system remains responsive. With chrome, it brings the whole OS to its knees, swapping and thrashing. And, yah my desktop machines all have between 8 and 24GB of RAM, but they also are running other apps that consume a fair amount of it (vmware for example).
need massive amounts of HDD I/O at all? And how did this even get marked "insightful"?
Its called paging, and it happens when you run out of ram, this could be because you have other apps open also consuming gigs of RAM. Its also particularly problematic because chrome is susceptible to javascript memory leaks like FF is. Actually, it might be worse in a couple cases in chrome (direct experience trying to fix a memory leak in one of our web apps). There is nothing like showing up in the morning at work, and your machine is crawling because your browser as gone and consumed all available RAM due to a couple leaky ajax handlers.
Your using chrome as an example? Sure its fast, but its also 6M LOC. The main reason apple chose KHTML/KJS was that it was easy to maintain at something like 150K LOC.
The problem is who are they going to replace him with, look at what has being going on at HP for a decade. That is another blown tech story, and fundamentally reflects what happens when your management is out of touch in the tech industry. You might be able to sell shitty cars for decades before anyone wakes up and discovers your product and company is total shit (GM!), but in the tech industry it happens much faster.
If civilian companies have them.
Especially if those companies don't have any military ties.
This is the same reason my GPU isn't a military grade device.