Slashdot Mirror
Looking Forward to Intel's Grantsdale and Alderwood
VL writes "Over the next several days, you'll be hearing a lot about Intel's significant upgrade to the Pentium 4 platform. Soon enough, that brand new Canterwood board you have will be yesterday's news as two new words will be on the lips of all enthusiasts... Grantsdale and Alderwood."
http://www.anandtech.com/printarticle.html?i=2088
Very weak, Athlon FX 53 thrashes a 3.6GHz Prescott on i925 in gaming, and simply beats it in a lot of other areas.
... upgrading from a 533Mhz FSB P4 system (2 Ghz cpu, not sure of chipset) and wait for this? Then (after this is obsolete) I need to buy a new board to switch over to BTX format? Ughh, no thanks Intel.
All this new technology, and mobos still have parallel and serial ports.
Get with the times!
-Apple
If history shows anything, it's that people who aren't gamers just don't really care too much about upgrading any more. Intel is going to have to raise its prices as sales due to upgrades slow dramatically. I'm still running mostly Pentium 2's in my business... I think. I don't even know or care. For what we do here, just about any computer that was made in the last 10 years is just fine. When it's time to get a new machine, we always just buy the cheapest oen we can find.
Am I missing something from the pictures there, or are these chipsets just a PCI-X + DDR2 update?
I just read the article, and it didn't talk about any major architecture changes in the P 4 -- just that Intel was integrating the latest and greatest in shiny new things into the motherboard (i.e. comes with DDR2 instead of DDR, PCI Express instead of PCI, etc.). Are these upgrades actually going to do anything revolutionary to the Pentium chips? Or do we have to wait until the Pentium 5 because all the changes they made are about compatability to the new technologies used?
Parallel and serial ports are nice to have, especially if you want to build some of your own hardware. And considering how insanely cheap a uart is, why not?
In fact I can invisage a day when most motherboards have inbuilt CPUs like they have inbuilt chipsets.
Beings aspergers AND pulling chicks... I enjoy the challenge!
I am not a fanatic when it comes to processors (i.e. not a die hard fan of AMD) but Intel is a rip off with respect to price compared to AMD.
Last few computers I bought were AMDs, I was very satisified with the cost and the preformance.
3dinfo@maficstudios.com
If your looking for revolutionary (or at least seriously evolutionary) advancements in chip design and architecture you might want to take a look at some new chips from a smaller company by the name of AMD. AMD's new Opteron and Athlon chips sport their new AMD64 bit instruction set as well as integrated memory controllers, Hypertransport interconnects and a NUMA style architecture.
Take with a THG Pinch Of Salt
9 /s ocket_775-15.html
http://www.tomshardware.com/motherboard/2004061
(yes, that is page 15 to start the chipset talk, there's plenty of stuff before that of course, but this is a chipset story)
...BIOS support for USB keyboards and mice has been standard for quite a while now. I've used a USB keyboard on my PC to make changes in BIOS for quite some time.
What is RAID 0+1 on 2 drives? Isn't that just a RAID 1 or a RAID 0 array? That doesn't make any sense. Yes, you could partition the drives into halves, THEN do a raid 10 or raid 0+1, but that defeats the purpose of reliability across multiple devices. If you have a hardware failure, you could lose both partitions.
-- Having a Creationist Museum is like having an Atheist place of worship
95% of the population has no use for legacy ports any more. In the future if people really, truly need legacy ports (i.e. no alternatives exist) they'll be willing to pay extra.
As for the USB keyboard/mouse issue. I'm able to boot into and use Open Firmware using my Bluetooth keyboard on my Mac. Maybe it's time to modernize.
That is just like saying "Hey, that new MoBo has PCI instead of ISA, WHAT IS THE BIG DEAL?"
PCIe is the future of PC internal AND external interconnects.
Just when that EFI firmware thing would make a serial console possible.
I rarely criticize things I don't care about.
Alderwood is a wood that, when burned, produces an aromatic smoke typically used for flavoring food. You can buy sacks of the stuff at Home Despot (so called because the manager of my local one is a tyrant) to put on the grill next time you barbecue.
To me, Alderwood seems an unfortunate name for a chip. I don't think it's a good marketing decision to name a chip for a wood prized for its smoking ability. That seems to evoke images of chips overheating and melting down in a puff of smoke.
Just take everything out of your box, and throw away your alim, memory, graphic card, motherboard, and so, what's left? The good old floppy drive, and the case! That is a little bit hard to swallow! Moreover, Intel cannot change all the current technology on its own: now, AMD is a serious alternative, and, thanks to the Itanium (1 & 2) saga on the server market, we all know that Intel's choice have to be debatted, moreover Athlon 64 FX are very impressive, and allow you to keep your hardware! I don't know why this article deals with PCI-X and audio chipset, it is not a CPU feature, but depends on a motherboard's chipsets... Definitely, my next computer will remain AMD powered!
From A problem with cinematic rendering on a VPU Where do the frames go? some other applications might benefit from it (examples given in the article). Although the author does point out that for AGP it is more of a drivers problem than hardware.
Looking at benchmark tests for these new mobos from Intel, one realizes how little advantage there is in all these new standards. There is nothing to be gained in extra performance if your current system is enough for your applications. And for those programs where more performance is needed, the overall system must be carefully balanced in order to give any appreciable improvement. The question to be answered is where is the bottleneck. If our eyes aren't capable of seeing more than 30 frames/second, as Hollywood has been demonstrating for the last century, then what's the use in a 100 fps graphics card? Why not swap fps for more triangles in a scene? Why not swap better graphics for better AI in games?
My own ideal system would have:
1) The fastest possible CPU, in *true* GHz, not in AMD's inflated "+" bogoghz.
2) Enough cache for at least 8192 single-precision floating point numbers.
3) At least 1 Gbyte RAM.
That's all. Those are the main points where current systems are underpowered. I mean, of course, for some applications. For text editing, the 4.77 MHz CPU in the first IBM-PC was ample. Looking at applications where performance in PCs still lags behind, I think simulation of physical systems is where most improvement is needed right now. Not just for "serious" applications, but also for games. It'd be nice to have a game with realistic simulations, like turbulence (think of surfing) for instance. What's the point in a 100 fps system if there's no way to simulate a wave breaking at the beach?
Another field where more progress is needed is artificial intelligence, but that's one point where software is less developed. One could simulate turbulence in a fluid rather well on a desktop, if only the hardware was fast enough, but no one is really sure on how to simulate the intelligence of a self-conscious being. But, anyhow, faster hardware wouldn't hurt either. If enough developers had systems with the same hardware capability as a human brain, I guess the needed software would be developed, sooner or later.
According to at least one tester. The higher latency overwhelms the bandwidth advantage. Given that AMD already had a big latency advantage with their 64-bit chips and the higher cost of DDR2, I don't see the big deal. Pushing DDR2 isn't as bad as pushing RDRAM, but...
RAID? That's nice, just about every high-end AMD board has a SATA RAID controller from Promise, Silicon Image, etc.
The audio is kinda neat, if there are Linux drivers. I doubt it's as good as a proper card but you can't argue with the price.
Anyone who buys Intel's "Extreme" integrated graphics to play current games is in for an extreme disappointment.
Wireless? (Cough!)...
On balance, all this hype over a chipset translates into Intel shouting "Pay no attention to our inferior CPUs!"...
We're talking about motherboard chipsets here, not CPUs. While looking at CPU architecture, clock speeds, etc. etc. to get a gist of how a PC will perform, it's still important to remember that speed of a PC is about the sum of its parts.
So think of these changes as an incremental speed increase across the Intel platform. Sure, they're a heck of a lot more boring than seat-of-pants GHz updates, but I welcome decent integration of a whole new set of bus technologies (SATA and PCI Express) which we've heard a lot of, but not seen much action on. Remember that PCI has been around for 10 years or so now and is getting a little long in the tooth stuck at a 33MHz bus speed.
In any case, it'll be interesting to see how these architecture updates are carried across to the Intel mobile platform.
Someone in Intel's engineering department needs to research the active heatsink for a chipset.
Look at the size of that thing.
Learn something new.
I'm with you, Captain. I still have a number of PII-450 servers (Proliant 1600), some of them dualies, that are as reliable as the sunrise and not coming anywhere close to bogging down on CPU utilization. And they're doing lots of work for us, too. I went recently to eBay and picked up some new power supplies and case fans for these units. I found those, and some hot-swap drives, too, at prices so low it was almost embarassing. I have a feeling these babies are going to keep producing for us for a long time. Recently, an NT4 Proliant became a RHEL 3.0 webserver. We gave it a new PS, a new case fan, and a full load of memory, and it's just cranking away for us. Gotta love it.
It's only funny until someone gets hurt. Then, it's hilarious.
Well, at least if the chips suck, I can build miniature guitars out of them....
In Soviet Russia, Chuck Norris will still kick your ass.
It's exactly as you say.
You have two drives, split each in half, and make two arrays with it, RAID 0 and RAID 1. Now, one of your disks dies. The RAID 1 part still works, because you have a disk left. The RAID 0 is dead, but hopefully you didn't use it for anything important anyway.
This way you can both have high speed and reliability with just two drives.
A PC in every room, except the bathroom...
Maybe Intel is just trying to save some room for growth for after every other room in the house has a PC.
Also, many projects rely on the cypress EZ-USB too, some of which even ended-up on sourceforge.
bundaegi is good for you
But could we at least make the product announcement more informative and less generic. I mean what use is it to say that Acme Unlimited is going to release Alderiumusian and Saphiriamius later today and all you Anaracrium whatzits are going to get you laughed at on the golf course. So if you want some action, upgrade today.
We are a tech board. We want to know what the upgrades are. What makes it cool. We are not reading Marie Claire in which the most important thing is that some pop singer has a new fragrance, or Fortune, in which the most important things is that some analyst was bribed to recommend a stock. I mean really, this post used a couple column inches and relayed nearly zero information except for a link.
"She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
The new P4's have 775 pins, the old ones have 478 pins.
What are those 297 extra pins used for?
Its not that Intel will go away anytime soon, but AMD appears to be ahead, certainly with 64bit (amd64) processors and perhaps even with its 32bit offerings. Intel seems to play down the 64-bit processors, perhaps because Microsoft won't have a true 64-bit OS for many years to come.
In the Unix world, we've had 64-bit OS's for many years running on SPARC, alpha and now amd64. My "64-bit future" started over ten years ago! There is certainly a 32-bit market created largely by M$, but M$ and 32-bit systems are past their prime. If I was Intel, I'd push the 64-bit hardware no matter how loud M$ cries foul.
It certainly seems, IMO, that AMD sees Unix as the future and produces far more compatible products. The Taiwanese motherbord makers should realise this too and stop fooling themselves. I'd gladly pay double for a mobo with quality features and less non-sence. Asus already seems to be doing this. The new (fairly low-cost 32-bit) A7V600 is a good example. It didn't take long to get all features, and more, useful or otherwise, to work under FreeBSD. (Even works well with 1.5GB RAM @ 400MHz while a maximum of 1GB is supported, presumably for Windows.) The Gigabyte GA7N-400 was an expensive disaster; Windows this and Windows that. I looks like it could work well with Linux, 400MHz RAM and a athlonXP-3200+.
I use computers for mathematical and logical pursuits. A "power user" in otherwords. I'm not impressed with gaming and 'cheap' polygon rendering. It takes a computing power of a true sort to produce holograms, stronger crypto, and related calculation intensive results. I do use a dual-Xenon, but its been a chore to tame. It was given to me with Win-XP installed! Linux-2.6.x seems very promising and FreeBSD-5.x might even be better? While all this is high-end equipment, its worth noting that Linux on a athlon-1200 is much faster (upto 10x) than Win-XP on the dual-Xenon! If people could only realise what they already have.
In closing, I don't see allot of merrit in using the latest Intel systems. The amd64 (Opteron/Athlon64-FX) will be the fastest thing on the affordable market for some time to come.
It looks like Intel is coming out with some compelling technology that addresses the major weaknesses and limitations of current motherboard and peripheral technologies. AMD has grabbed (and will retain for some time) a lead in pure processor performance, but overall system performance (as perceived by the user) and the overall user experience is built on more than just how fast the CPU is.
So, my question to those who follow this industry closer than I do is how will AMD position itself for success? Will motherboard manufacturers come out with AMD-compatible boards that sport PCI-Express and the other (non-CPU) new features that are talked about in this article? Or does AMD have another plan?
The ECC logic is broken on the current stepping of the Alderwood chipset.
Mea navis aericumbens anguillis abundat
why do I never have mod points when there is something amusing to moderate?
Panurge has posted for the last time. Thanks for the positive moderations.
They Sound like the names of multi-million dollar gated communities...
Can you ping me now? Gooood! | Manhappenin.Net - Things to do
When VESA came out, I had to get rid of my ISA video card. When PCI came out I had to get rid of my VESA card. When AGP came out I had to get rid of my PCI card. When PCI-E comes out I have to get rid of my AGP card. So? Why is the PCI-E move any worse than the move to AGP?
Don't get me started on the different types of memory which I've had over the years. But, I wouldn't sit around arguing that I was screwed over by the move to DDR, for example.
PCI-E paves the way for much higher network bandwidth, more bandwidth for graphics cards, etc. PCI-E will scale to at least 10 GigE, if not beyond. Some of this means more in the server room than on the desktop, but it's nice to see the bar significantly raised across the board.
I recall reading somewhere that some motherboards would probably ship with AGP slots as well (AGP->PCI-E bridge?). Legacy PCI slots will also be available on many/most boards. You don't have to buy the board which supports DDR2, so you should be able to use your existing DDR memory. So, you need a new motherboard, CPU and case and can then grow into the rest of the new technology which is offered on the board.
I doubt you'll hold the same opinion several years from now. I think you'll look back and see that this was a good move, just as moving to PCI was a good thing, moving to AGP was a good thing...
In my post, which you obviously didn't read, I point that two types of applications where even the fastest PC CPUs today are lacking in performance are physical systems simulations and AI. Let's see why.
Imagine a typical simulation, for example in a 1000x1000x1000 box. You have one billion points for which you want to calculate the evolution of some physical measure, let's say air pressure. For each iteration of your software, you need to do one billion mathematical operations, usually sums and multiplications. Of course, you'll want to display the results, you want to have some kind of user interface, etc. But all this is of little significance, compared to the task of doing some billions of floating point calculations.
So, my point was this: if you *really* need that CPU, you need to do lots of floating point operations per second. Yes, before you mention it, I know there are other types of software. But for anything other than doing floating point operations, the current PCs are ample for any personal software I can think of.
In the end, it's the mathematical calculations that slow down a CPU, not the logic in the software. CPU developers, at Intel or AMD, go to great lengths to optimize things like function calls, loops, tests, stack operations, etc, but it's not necessary. With the current CPU speeds, logic optimization needs not be taken any further. It's the floating point operations that get most of the processing effort. And those have been optimized to the last level. Pentium, or AMD, or PPC CPUs, they all can do an addition AND a multiplication in a vector of four floating point numbers in one CPU cycle. The only way to improve that would be to increas the size of that vector, but that would increase chip size (and cost) and power dissipation.
AMD may say what they want, but CPU speed IS the main factor in performance. Because, in the AMD formula above, the "work per clock cycle" is the same for each manufacturer. I mean for those programs where the CPU is really not fast enough, those programs which you start running now and come back in a couple of hours or in a couple of days. If I get a 10% improvement in a program that runs for two days, I gain 4 hours and 48 minutes in each run. Much better than getting 10 microseconds less for each spell checking.
I still use the PS2 ports for my mouse and keyboard on the PC I built earlier this year. Simple input devices simply don't need the bandwidth of USB 2.0, and since I only have 6 USB slots available, and five of them used (printer, joystick, CF/MD/SD card reader (uses 2), PDA dock) full-time already, I don't have much left for another device unless I bought a USB card.
Though, I'd wouldn't mind dropping the PS/2 support (and serial and parallel) if mobo manufacturers replaced them with extra "new" ports, like USB 2.0 or firewire and other companies (like Logitech) quit making PS/2 devices.
Sigs are for losers
PCI Express isn't as big a jump as it sounds like. The new Dell Poweredges have the ServerWorks GE bus architecture, which uses five separate PCI busses of various widths and speeds. This puts very few items on any given PCI bus, and PCI Express is just going to mandate one device on any given connection. I'm sure other manufacturers use similar technology.
A good news: you can put a PCI card in a PCI-X slot, the counter part, is that the whole PCI-X bus will act as a standard PCI one...
Moreover, on the graphic cards use, 4x is already enough, 8x gives a good margin, but, I don't see any need to increase graphics card bandwidth? Screen resolutions won't change tomorrow, and framerate improvements? I don't see the point, graphic cards already have their own memory, often faster than the one dedicated to the rest of the system... May be cool for huge graphic wall, with plenty of cards in one box? But ok, clusters dedicated to graphics already do that.
There are already Tyan motherboards for Xeon and opterons that give both PCI-X and AGP, that will only depend on Intel's will to ship chips alowing that, and, as I have seen it through Itanium, Intel like to discard technologies (for Itanium, they have even changed the standard for bootable CDs!!), this can be seen through the new BTX alim (BTX... why not ATY or AUX??). That is why I am so reluctant about this anouncement: you don't need PCI-X to enjoy broadband Internet, you don't need it for 3D apps, as well as you did not need a P4 to "surf on the web" as they claimed it on their ads... I have never liked Intel strategies to own the market!
This is only half the story. I feel the change from IA32 to AMD64 instruction sets is equally significant. It's a shame Intel won't just bring out the entire platform at once, since many people buying their 32-bit desktops with these new support chips over the next few months may very well feel their systems were quickly obsoleted when the new instruction set ships.
And while it's only my opinion (lawyers take note), I feel Microsoft is colluding with Intel by not releasing Windows64 until Intel can be fully caught up with AMD's lead. They had good versions of Win64 running many months before the first Opteron hit the market last September, and it's still not released!
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
You sound like a shill for Intel.
Or that clockrate (gigahertz numbers to those of you in Rio Linda) is only half the story. The other half is the number of instructions executed per clock. In this regard, AMD Athlons are more efficient than Intel Pentium 4 chips because each AMD chip does more per clock tick than the corresponding Intel chip. That's why AMD can do as much real work as Intel at a lower clockrate.
And if you feel, for some completely unexplainable reason, that this idea is bogus, well Intel has gone back to it with their Pentium-M chips. These P-M chips, at clock rates of around 1.6GHz, do as much work as Pentium 4 chips at a Gigahertz higher.
Either way, your attitude is what keeps Intel profitable -- and you poorer than you might have to be if you were a more savvy buyer.
By the way, when you buy cars, to you look at the speedometers and pick the one that reads to the highest value? Sorry, just had to ask.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Major audio improvement over AC97 at essentially no additional cost.
8 USB 2.0 high speed connections.
Firewire? (The Intel manufactured boards at least are including this.)
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Except that's ridiculous. It can't possibly work since all data must be stored on both disks somewhere. Writes must go to both disks while reads could be done from either. You get that with simple mirroring in the first place.
They did. It's not new and it's a sham.
As if that would work. After RTFA try turning on your brain.
For you to make this arrangement work, you would have to split the drives into halves, create a RAID 0 array on each have, then mirror one array with the other. In addition, you would reverse the role of the drives in the two arrays so that mirrored blocks would always reside on opposite splindles. Then, to avoid large seeks you would strip partition 0, drive 0 with partition 1, drive 1 and likewise 1,0 with 0,1. For all this fancy manipulation what you end up with is EXACTLY RAID 1. Why? Because it's impossible to do both RAID 0 and RAID simultaneously with just two spindles. Writes must go to both while reads can come from either. Just like RAID 1.
Or something akin to that? Granted I only gave the article a quick read through, but it looks like the only real difference is overall base bus speed. 20 year old IBM technology strikes again.
Does anyone else think that it's about time for Intel to retire the "Pentium" monicker? I, for one, find that it brings forth connotations of "overly expensive" and "marketing-inflated", and the likely desired effect is lost in today's world where AMD stuff is performance, and Intel is trailing. Granted, Intel has better low-heat and low-power chips, but most folks don't even see that as a potential feature to be considered, let alone a feature.
On the other hand, Intel has probably spent trillions of dollars advertising "Pentium" via the brand name and the little tin-foil-clad men (and TBMG). Have they over-extended their brand name to a point where it's entered into common acceptance (ie, "google" or "tuperware")?
What's everyone else think? Is "Pentium" old and tired, or does it still have viable commercial life left? How about as appeal to geeks (as we all know that geeks are the ones that push the next industry trend into the forefront)? My personal opinion is that "Pentium" is outliving any usefulness it might have, and doesn't really have the "whiz bang super-fast" association that it used to (there are Opterons, Xeons, and god knows what else now, after all), at least with the geeks. Also, I think Intel could benefit from a newly marketed name (provided they have the chip to back it up - maybe a low-power portable chip?) to blow away the industry again, leading to the trend of further portable device purchases (supplanting desktops with laptops, for instance).
~/ssh slashdot.org ssh: connect to host slashdot.org port 22: too many beers
>Look at the size of that thing.
/Overclock? In our moment of triumph?!
Cut the chatter, Red Two. Accelerate to attack speed.
Can't stop the Beta? Time to evacuate to ##altslashdot at webchat.freenode.net - Slashcott in effect.
I have been holding off on buying a new computer for almost a year now. Waiting for a few key items to fall into place. My current rig is a Athlon 1.4ghz, 512ddr, 40gb c:, 200gb d:, Geforce 4Ti. This has worked well for me for almost 3 years now. I really see no need to upgrade except to run games at higher resolution.
My wish list is:
DDR2
Gigabit Ethernet
3.0 Ghz Intel (I dig hyper theading)
SATA2
SATA2 can do Command Queueing to speed up data retrival. This is a big thing for me as I see this new rig will last me till 2008. When I do upgrade my hard drive in a year or so I can get a 10,000rpm SATA2 drive.
Does anyone know any details is 915 or 925 will have SATA2?
PCI-X is not PCI-E (or PCI-Express). PCI-E is a completely new interface and is not backwards-compatible like PCI-X.
To read the specs on both, click right here.
Sigs are for losers
I'll bet the cost of those big connectors, board space, and punch-forming metal openings for the ports add up.
Still, you're right, the UART is cheap -- perhaps they should consider keeping it, and just put cheapo internal-style multi-pin connectors on the motherboard. People who want to use the ports can add adapter cables to the old-fashined connectors.