Slashdot Mirror
Intel Releasing PIII Xeon Today
BMIComp writes "Yahoo! news is reporting that Intel is going to introduce their Pentium III Xeon Chip, today. " .18 microns, 700 Mhz, and integrated cache. The article talks quite a bit about how the new Xeons are going straight for Sun's throat.
Maybe you should go see a doctor about that... :o)
Got Rhinos?
Getting into issues of RISC versus CISC gets us away from the point.
The point is that those that are preferring Sun systems to Intel-based systems today are doing so not out of being focused on how much stronger SPARC used to be for CPU-intensive processes, but rather (most likely) because they've got I/O-bound processes.
The result is that the announcement doesn't scare Sun:
But as there weren't Origin/Crays running SPARCs (I don't think there...), while this may represent a threat to MIPS or Alpha, it's not particularly one to SPARC.
Overall point: The article suggests that the threat was to Sun. A "threat tree" shows that this is not strongly the case.
If you're not part of the solution, you're part of the precipitate.
I tried to post this with this news story. It's on C|NET. They both sat pretty much the same thing.
Devil Ducky
Devil Ducky
MY peers would get out of jury duty.
tcd004
Here are my Microsoft and AICN parodies, where are yours?
Remember the Celeron 300 and 300a? Case in point. The more the better. Your logic is flawed here. It's true that going from the cacheless 300 to the 128k cache 300a was a huge improvement I wouldn't say "the more the better." It depends entirely on what application you are using the proc for.
For instance, a computer with 32MB of ram is going to run a ton faster than one with 8mb of ram, but one with 1GB of ram isn't going to run that much faster than one with 256MB of ram for normal applications. Only really high-end apps need this much.
The Xeon is aimed at High-end servers which could probably benefit from the cache, but is the price worth it? I dunno.
The main reason to buy one, as you stated, is for the >2 SMP. AFAIK the only difference between the MP capabilities of the Xeon line and the regular line is that the PIII's are purposely crippled to only run on 2x SMP. It seems like a rip that Intel would do this and then charge an arm and a leg for almost the same proc with SMP turned on.
http://overwhelmed.org
Ushers will eat latecomers.
IP is just rude.
Is there any torture so subl
As far as I can tell, the most the Xeon can scale to is 8 processors.
And it doesn't do that well. Intel's 8 way chipset is crap. To grossly oversimplify it's two 4 way shared busses communicating over yet another shared bus. It performs about how you'd expect. Compaq has a better 8 way xeon chipset, but if they're not your hardware vendor, you're SOL.
Somewhere Scot McNeally is screaming "Help, I'm being menaced by a dwarf!"
Or not.
--Shoeboy
(former microserf)
Sun's systems consistently underperform when matched up against similar systems from almost all of their competitors.
Depends on the jobs, benchmarks, etc, blah blah. In the market the EXX00 systems reside, they offer some damned good scalability and redundancy (predictive failure modeling on CPU/RAM and hot-spare CPU/RAM without system reboot.. Add new CPU/RAM as hot-plugs and activate without system reboot... How many Intel systems can do that?) as well as memory capacity and the 64-bit OS to support it.. I'm looking into refurb 3x00/4x00s for several system upgrades, and while some of the SPOFs are annoying (primary CPU/RAM planar errors require reboot, no alternate-pathing of PCI boards) I'm still interested in the overall improvement they offer.
Sun offers the best scalable unix systems for business use IMHO. Between CPU/RAM options, software availability, driver support, and OS features I'm pretty happy with it...
Now if Sun'd ever backport DSDs to the XX00 series like they did with Starfire's DR/AP... one can dream...
Your Working Boy,
Too bad that availability is still a joke, as well as the exorbitant prices. Way to go Intel!
What makes the Xeon better? The cache is larger, yes? Everyone else on Slashdot knocks L2 cache saying it's overrated and underutilized as is. Anyone have any insight as to what makes the Xeon a good choice in the server arena?
True, the cache is 'off-die', but it does run at cpu speed. This is why the Sparc modules have about 2 pounds worth of heatsink on them.
It is also why a 450MHz/8MB Ultra module costs $9500. (Note: I haven't looked in quite a while, so that price was when I last looked -Dec, 99-, it is likely less now, and I am sure big customers get deep discounts.....)
In the immortal words of Socrates, who said; 'I drank what?'
Three years ago the IA-64 was supposed to put Sun out of business. Since we are now unlikely to even see a broken IA-64 in the next year Intel has dusted off the losing Xeon and will continue to wage a war of words with it. Meanwhile Sun marches from strength to strength with its multiprocessor system and the solid SPARC chips. Intel, you could learn something about engineering from Sun.
AFAIK, Athlons *do* support SMP, it's just the motherboards are missing... Athlon is using Alpha EV6, which supports per processor bus in contrast of Intels shared bus between all processors. Intel has some major problems with >4-way systems just because of this. Athlons should be able to do better when we finally get those MBs. :)
Yep. I've read the whole thing. :) It's exceedingly easy to program as well. And has backward compatibility in such a way that doesn't interfere with the rest of the CPU. 44-bit (64 in US-3) address space. Memory-mapped I/O. I could go on... Intel could learn a lot from the Sparc.
I think that Intel is going to be stuck in the personal computer market for a while. They just don't have what it takes (yet) to play with the big boys.
Got Rhinos?
What makes the Xeon better? The cache is larger, yes? Everyone else on Slashdot knocks L2 cache saying it's overrated and underutilized as is.
For most personal systems, adding cache would give diminishing returns fairly quickly. However, a personal system usually has one or at most two major tasks draining resources. A server has many cpu- and memory-intensive tasks running at once. A small cache would thrash on every context switch. A larger one wouldn't. Of course, how much of a performance gain this gives depends on how long the timeslice is, but if the timeslice is short enough, it's relevant.
Anyone have any insight as to what makes the Xeon a good choice in the server arena?
Probably price. A Xeon system is cheap compared to its competition, if I understand correctly. It doesn't scale as well as a Sun box, and won't give you the floating-point performance of an Alpha box, and doesn't have the memory integration of an SGI box, but as an inexpensive mediocre solution it can be a good buy if you don't really need something cutting-edge.
(It'd prolly be faster, too)
Got Rhinos?
The new Xeons will be used in a 32-processor server from Unisys that Compaq also is selling,
.
Ambrose said
No, that's wrong, they will be used in the design of those systems, to figure out how to most efficiently dissipate heat. .
I just remembered this old Metallica song. . .
These are my friends, See how they glisten. See this one shine, how he smiles in the light.
"What's everyone else's opinion on this? "
um. cool? I wasn't going to be buying any Intel chips anyway.
I just remembered this old Metallica song. . .
These are my friends, See how they glisten. See this one shine, how he smiles in the light.
Besides being a patently dumb name (sounds like a new Noble gas), Xeon is a chip with split personalities.
The original PII Xeons were the standard Deschutes (.25 micron PII) cores with full-speed L2 cache at 512 KB, 1 MB, and 2 MB. Those were replaced by the PIII Xeons that had the Katmai (.25 micron PIII) core with the same cache and sporting the addition of SSE. Both were rated for up to 8-way SMP. These are the Xeons that maxed out at 550 MHz.
There are a newer batch of Xeons based on the Coppermine core (.18 micron) that don't really differ from today's PIII's except that they are rated for multiprocessing (2-way only, I think). The Coppermine Xeons have 256 KB of on-chip L2 cache, just like the Coppermine PIII's, and can run on the 100 MHz or 133 MHz GTL+ bus, just like the Coppermine PIII's.
Skip ahead to the last week. PIII's are now rated as SMP (2-way only) capable. The Xeons being announced have SSE and on-chip cache, but the cache is (mostly) the same size as the old Katmai Xeons, namely 1 MB and 2 MB. I guess 512 KB is gone for good. Also, the new high-speed Xeons are capable of 8-way SMP, like the old Katmai and Deschutes Xeons.
One interesting note on the stability and scalability of Intel's bus design (remember it's been in use for 5 years+) is that they have pushed a bus that started at 60 and 66 MHz to 133 MHz, but in order to allow SMP beyond 2-way they can't get above 100 MHz.
GK
Please note that I did this all from memory and with all the holes in my for eating and hearing and whatnot I'm surprised all my memory hasn't leaked out yet. Now warned, you may flame away.
like amiga users n such
I eat the flesh off the living, and I vote!
And how can these CISC processors outstrip Sun's RISCs? I don't care how many mhz it runs at, the two just work differently.
There are efficient ways to implement CISC; the most common is to break down CISC instructions into smaller RISC-ian operations that are easily pipelined. You might have an extra clock's latency for the decoding, or you might not, depending on architecture.
The functional units that perform operations are the same. The cache and memory subsystems are the same. Processors tend to be bound by design tradeoffs in these or in the motherboard or in things like the register file size as opposed to by the instruction set.
Short version: RISC and CISC are different ways of telling a processor to do the same things. CISC used to make scheduling harder. Now it doesn't.
D
----
Pentium II was just a PPro with MMX.
Will I retire or break 10K?
Sun's architecture is based on high-speed switches connecting multiple fast/wide PCI or Sbus buses. The peecee architecture is based on having one to four PCI buses on another shared bus. Suns have memory buses no less than 1152 bits wide. Peecees generally have 256-bit memory buses. Summary: Suns can move far, far more bits around inside them than any peecee server can. Sun 4, Intel 0. It should be noted that this isn't entirely Intel's fault, but they aren't really helping anyone do something non-peeceeish. You _could_ build a Sun-like system with Xeons, but nobody does.
The peecee architecture is hopelessly obsolete. 16 bit bootstrap code. A BIOS. Forget boot monitors and intelligent peripherals, you know, things that make systems manageable and flexible (just try booting a peecee from tape - you can't - or over the network - that'll cost you extra). Both system architectures have their roots in the late 70s and early 80s. Sun started with a good design and have steadily improved and enhanced it. The peecee started with a poor design and have left it essentially unchanged for 20 years. Sun 5, Intel 0. Again, not all Intel's fault, but since there aren't really any other systems available using Intel CPUs, it counts against them anyway.
Now, even with these drawbacks, I could see the Xeons being a good choice for a midrange server if they cost much less than the Suns. Unfortunately, the Xeons are actually more expensive than the UltraSparcs, even at the same clock and with the same size L2 cache. If you have $4k to blow, you can get either a Xeon 500 w/2MB, or an UltraSparc 450 with 4MB. And the UltraSparc is faster, too. Oops, guess that's 3. Sun 6, Intel 0.
Disclaimer: I used to work for Intel's server division, so I may be biased...
Yup. You see a noted difference when moving from an old Ultra-Narrow 5400 RPM drive to a brand new Fibre-Channel 10,000 RPM drive...
Those help, but RAID controller speed doesn't affect overall speed as much as you might think. The biggest speed improvements for RAID controllers are: bus improvements (new 64-bit controllers,) interface improvements (Ultra-160 or Fibre-Channel,) more cache (Yes, 128MB of cache helps ALOT in Database applications,) and, more channels (it really helps to have your drives distributed among a few channels. That's why three-channel RAID controllers are so good.)
Actually, the current limit on Intel servers is 64GB. One of the servers I supported when I worked at Intel was their OCPRF100. An eight-way behemoth that had 64 DIMM slots, each one capable of taking a 1GB PC-100 DIMM. The big problem is OS support. NT (including 2000 Adv. Server) only support 8GB (they claim more, but they're lying,) and most other high-end server OSes either peak at 8 or 16. 64-bit Linux doesn't have that limitation, but of course, it's only for use on the upcoming 'Itanium' processors. From what I've heard, Win64's limit will be 32GB, but that's just a rumor.
With current 64-bit, 66MHz PCI (528MB/s of bandwidth as opposed to 32/33 PCI's 133MB/s,) and the upcoming PCI-X (1GB/s+ bandwidth,) system bandwidth is improving. Combine that with gigabit ethernet, Ultra-320 and Ultra-640 SCSI, and in the next two years, bandwidth will no longer be an issue. (PCI-X, GigabitE, and Ultra-160+ all need I2O.)
I just miss playing Quake 3 on that eight-way system...
Another non-functioning site was "uncertainty.microsoft.com."
The purpose of that site was not known.
Although Sun support up to 64 CPUs with the E10000, they're not all on one motherboard. Similarly, 64 CPU Intel servers (such as the Data General AV25000) will typically only use 4 CPUs per motherboard. Some newer configurations may use 8, but time will tell. The difference is that Sun have 64 CPUs in traditional SMP configuration. The Intel machines tend to have them in NUMA configurations.
"The invisible and the non-existent look very much alike." -- Delos B. McKown
But it's not necessarily the truth.
The truth is that for a whole lot of the applications for which people are installing big UltraSPARC boxes, the applications are not CPU-bound, but rather I/O bound.
Having a better, faster Xeon Pentium III processor doesn't help with either of these things.
What helps with such applications are:
Which Seagate and Quantum and IBM are responsible for...
Think Adaptec, maybehaps?
Which means more than the 8GB that appears to be supported at this point.
If you're not part of the solution, you're part of the precipitate.
Now I know I could have the traditional beowolf cluster of these, but how many can I get in one box/board? As the article says Sun is winning in the big server market with upto 64 processors, what is Intel offering, are these 2/4/8 way limited processors, or are we just going to have to wait and see what sort of motherboards appear?
Never underestimate the dark side of the Source
(ASPs), a market also targeted by
Sun Microsystems Inc.
I find this very intriguing - why would Sun be intent on targeting the ASP market? You'd think that with their focus on Java, they'd be much keener on simply convincing that market space to switch to JSP's or servlets, both of which would offer far superior performance on Sun servers.
This seems very odd -- does anyone have any comments or explanations for this?
Keep in mind, Microsoft wants everyone to think DNS stands for "digital nervous system", too. Fuck that!
- A.P.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
Not to be abtuse here, but I don't get your analogy. Straight line speed is one thing, data throughput is another thing. So I thought maybe you meant that while Sparcs are slower that the PIII, they typically come with better I/O infrastructures. I'm all set to agree with you there. After all IBM sells big iron on this argument alone.
But then you said load, and I'm thinking maybe you meant straight line speed anyway.
So could you clarify?
I'll try and kill two birds with one stone.
jmv is more correct than I in that it isn't strictly the cpu which is designed to 'handle' the load. I should have been more clear in that regard.
What I should have said is that both the cpu and the OS are designed to handle higher loads more efficiently. With Sparc cpus, task switching is aided by the Context registers built into the silicon. I am no hardware designer, so I couldn't say with any authority how x86 handles task switching.
So, to answer one of the above respondents, by load I mean how many tasks a cpu is required to switch between, and how efficiently both the cpu architecture and the OS can do that.
Don't think 'speed' when I say 'load'. Rather compare the ability of the machine to respond when the 'loadavg' starts approaching 1 or greater per cpu.
In the immortal words of Socrates, who said; 'I drank what?'
change it to, "how the new Xeons nibble complacently upon Sun's ankles" :)
I got this information from a textbook, not from Intel documentation, so I may be a little off. Anyone know more about this?
As the article says, Pentium III Xeons have been around for a while, but were topped out at 550 Mhz. This is just faster (and different layout, I think).
Pronouncing Intel's chip names make is so hard it goes for *everyone's* throat. Just saying "xeon" makes me want to hurl.
--
Have Exchange users? Want to run Linux? Can't afford OpenMail?
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
well-it's out. 700MHz isn't really going to turn that many heads, though, now that people get glassy eyes when they hear 1GHz. but it's really not fair to compare the two. the fastest UltraSPARC right now is what, 440MHz? again, it's not fair to compare that to a GHz p3. I think Intel and Sun's marketing departments need to make the distinction clear between their processors. I have seen people compare my 400MHz Ultra 10 at school by saying "oh, it's only 400mhz. i have a celeron 450, how much better is it than that sun?" now we all know that a 400MHz UltraSPARC flies. it's an incredibly fast processor. as such, sun is justified for charging enormous amounts. it's not the megahertz that matter. as such, i still wonder why intel would flaunt the fact that the new xeon is 700MHz, but yet advertise their GHz pentium. i say long live the 200mhz pentium pro.
What's everyone else's opinion on this?
Got Rhinos?
Intel might claim that they are going for the throat with the PIII Xeon, but they have a huge gap to close before marketing catches up with reality.
...same as a standard PIII.
As far as I can tell, the most the Xeon can scale to is 8 processors. At least that is the largest machine I can find for Xeon. Sun's midrange machines _start_ at 4 processors and, for now, go up to 64. The planned maximum for UltraSparc III is 1024. Sun 1, Intel 0.
UltraSparc is 64 bit, Xeon is 32.
Sun 2, Intel 0
UltraSparcs have had integrated caches for quite a while, as far back as SuperSparc I. On the current cpus, the desktop boxes can have up 2MB cache, the midrange servers 4MB, and the large Enterprise servers get 8MB cache. The PIII Xeon's claim to fame is the extra electronics they add to the processor card which allows some hardware admin of the processor. The cache size is 256KB
Sun 3, Intel 0.
Until Intel produces a cpu which can scale well and has an OS which supports that scaling, I don't think Sun has much to worry about.
And for those of you who think MHz is a consideration, bzzzt! In the world of 24x7 server farms, torque counts more than horsepower. The UltraSparc cpu can handle a much greater load without sweat than most other cpus. The UltraSparc was designed to handle massive loads. So, while the UltraSparc will likely lag behind an equivalently rated Intel cpu, which would you rather use to make a cross country houshold move: A Kenworth or a few dozen pick-up trucks?
In the immortal words of Socrates, who said; 'I drank what?'