This Benchmark is totally unfair!!!!! The PC-DL has the avalibility to produce 4 usable CPUs...2 Physical and 2 Logical.... Windows XP will only see 2 of these. The benchmarks are being done with Half of the POWER POTENTIAL.
I am having a problem with my PC-DL board. The System Management BIOS is reporting that there is 4096 MB of RAM, but Windows reports that 2048 MB is available. Anybody ideas? I am trying to get help from Asus tech support but any help would be appreciated. Running dual 3.06 with 4 x 1GB sticks.
I totally agree with FutureShock999. Why not run several instances of the a divx encoder. For instance do a 2 part movie on an encoder that runs two threads per instance then run both of the parts at the same time in two instances of the program on the xeon and see who finishes first. What you guys are doing is like comparing an f-16 fighter jet to a f-15 and saying the f-15 can only use one of its engines.
I got my pc-dl with 2.8GHz Xeons with raptor drives in RAID0 to 3361MHz 160FSB with a dram clock of 200MHz. 4:5 It ran Sandra burnin overnight. Maybe my board is a newer revision. I read this article after I bought the Raptor so I didn't try to overclock it until yeaterday but it did and it runs well!!!!!
Wesley, A nice review, but I believe the wrong benchmarks. No one should buy a dual-processor machine to execute a single-threaded application faster, especially when that single proc is slower than others on the market. Dual-proc boards are bought to do multi-threaded stuff, either runnig a single multi-threaded application, or running several different single-threaded applications. In no place did I see benchmarks that explicitly looked like that.
As such, your review was a good cautionary tale for people that didn't KNOW the above, and hopefully will stop some people from spending a lot of money on a this board hoping to have a great UT setup.
Now if you had shown what it was like to play a game, encode media, and download a few gigs of content SIMULTANEOUSLY, then we could really see how this board stacks up to the competition you evaluated it against. And I think it might have beaten them rather handily...
#23 - We recently changed our standard video card from nVidia's Ti4600 to the ATI RADEON 9800 PRO. Theoretically this should have no impact on encoding scores, but we reran all benchmarks with our new standard hadware on a few of the highest performing boards. Evan ran about half the new encoding benchmarks on the west coast, and I ran the other half on the east coast. As you can see our new results compare very well to each other.
I have no other explanation, but perhaps Evan can shed some light on this. I have used the new ATI Radeon 9800 PRO from day 1 and my benchmarks have been cumulative over the last couple of months with no dramatic change that you point out.
The 865/875 P4 boards in this test all perform 50% faster in the media encoding benchmark in this review than they did in the round-up a couple weeks ago.
Has anyone else noticed this dicrepancy? The Abit IS7-G has gone from 64.45 fps to 104 fps. Both times they used a P4 3.0Ghz, 800mhz FSB, with HT enabled. Nothing seems to have changed except the encoding speed. I wish I could do that to my rig ; )
A 50% increase in as linear and consistent a benchmark as DivX encoding is simply astounding.
I just wanted to point this out to everybody around here.
It' will not overclock with the SATA drives on the intel controller. BUT it will if you run them on the onboard promise controller. I have 2 WD Raptor's running RAID0 and the 2800/533's running @3250 100% stable.
Which brings up the question about whether or not HT was enabled on it. And the fact is they used a regular consumer card for video Most workstations would have a workstation class card in them such as a Quadro, FireGL, or even a 3DLabs card in them. It all depends on the applications that one uses. Most normal people wouldn't use a dual machine for gaming anyways. They'd use them for graphical processing or media encoding or file serving and such. Just talk to those guys over at 2CPU.com they know what I'm talking about. ;)
From the looks of it, there would be little if any reason to spend gobs of extra money on a system that is beat by AMD in gaming, and by single P4 siblings in high-end workstation tasks.
Aren't the Xenons at 667FSB? But as usual, they share the bandwidth, which is really bad since they crave bandwidth, but good for memory access. But clearly the HT idea is faster.
Running though at 800 maybe really too hard to do, or not ?
Remember Xeons were built for the server market where stability is king. Intel knew this and this is why the 533MHz FSB has stuck around so long. As Intel rolls out more 800Mhz FSB chipsets and P4 chips the 800Mhz FSB Xeon should follow several months later. The same process could be seen back when the 400Mhz FSB Xeons were around.
Well don't take my word for it, wait for the results yourself and then you can admit you are a fool! Clueless kids looking to argue... What a waste of time.
@13 "it would not improve performance enough to even be competitive"
lol... what reviews have you been reading? the extra fsb makes a huge difference! THG overclocked the Xeons by 7 mhz and averaged somewhere around 4-4.5% performance increase! That's pretty damn significant. Even bringing the Xeons up to 667 FSB would make it a very good performer!
Don't believe everything you read... ESPECIALLY from Intel. Go read THG's review and *maybe* you'll have a better appreciation for why Intel ain't likely to release a 800 MHS FSB Xeon any time soon and it would not improve performance enough to even be competitive when a 2.0 Gig. Opteron/A64 blows the doors off a dual 3.06 Gig. Xeon. If you wanna feel bad... look at how the dual Opteron beats up on a dual Xeon.
Read the news -- Intel has said it WILL produce 800Mhz FSB Xeons (when is a bit of a question: the last report I read over a month ago said Q1 '04 but they may have moved it up -- that's what this article seems to imply).
There are technical reasons why the Xeon can't do 800 MHz FSB. Tom's Hardware went into some of the reasons in their dual Xeon test where Opteron also smoked the Xeon even with the new L3 cache. The reality is that Intels current processors have just become obsolete with Opteron and A64. The benches for A64 will show equal or better performance for a single CPU than Opteron. For multi-processor systems above (4) CPUs the Opteron has an advantage over the A64. As the benches show Intel simply has nothing to compete in either the desktop or server segments and soon the A64 will be available in laptop to so Intel needs to get it in gear. Needless to say all Opteron/A64 CPUs will run both 32-bit and 64-bit O/S's. Linux has been available and Windoze will be available soon. AMD has definitety made a qunatum performance leap over Intel this time, no doubt.
Remember that server boards have to go through more stringent evaluation then desktop boards. I don't think Intel even considers 800Mhz FSB stable enough for their Xeons yet, and it's a shame. There is only so much increasing your L2 cache size can do for you. I don't think you can use 2.4Cs on this board even if you could find a socket adapter -- remember the 'overclocking settings' were limited to 133->165 (that's almost the 667Mhz FSB that Intel might move the Xeon to soon). So you'd be underclocking it by at least 17%, and at that point why bother? That's if you can find some video card that will take a 79Mhz AGP speed -- good luck with that.
This review makes, me wonder...is it possible to use two P4 2400/800 CPU's on this board, maybe with some kind of adapter. Those P4's would likely outperform this Xeon setup and possibly also the Opteron, with a very interesting price tag, even if these adaptors would cost $50 each.
And if anyone should see an Intel engineer, just tell him we want a 1000 MHZ FSB for those Xeons, as the i875 would surely pass all validation test at this speed, and the PC4000 memory is already available.
"we are reviewing the PC-DL, and comparing it to the performance from the current top Pentium 4, Barton, and Opteron/Athlon64 boards that we have tested"
Performance Test Configuration Processor(s): Intel Dual Xeon 3.06 (1 Mb Cache) 533FSB AMD Opteron Socket 940 at 2.0GHz (9x222) 444FSB Intel Pentium 4 at 3.0GHz (800FSB)
No Barton in there, the top Opteron (or so I thought) was the 1.8GHz, and the top P4 is the 3.2 as far as I'm aware.
Dual Xeons is a joke. The P4 is starved for bandwidth in a SINGLE processor configuration at 533Mhz FSB. What makes anyone think TWO processors can handle being that starved? It's like eating spaghetti through a straw -- you can get a few noodles, but it's faster to go buy a fork. And, last I heard, Intel wasn't planning on bring a 800Mhz FSB to the Xeon, just a stopgap 667Mhz FSB. Almost any task can be performed better by a single P4 3.0Ghz processor then dual Xeons of any speed on any platform. Xeons make no sense as a consumer platform all. And, with Opteron here, they make no sense as a server platform, either.
When can I see my Athlon64 vs. Pentium 4 benchies? I hardly think an overclocked Opteron counts. After all, the 2.0Ghz Athlon64 isn't going to run with dual-channel memory, anyway. Can you say Socket 754? Luckily, AMD is already moving to Socket 940/939 for future releases. Mmm. 64-bit goodness.
I'm always dissapointed in dual cpu system reviews, as they never show how the systems multitask. I would like to see the FPS in UT3 while encoding a movie at the same time. Can you play games while burning a DVD? I would like to see the performance while performing multiple tasks at the same time.
Yea, I'll bet Intel thought they could Spin dual Xeons as a viable filler until Prescott arrives some day. Reality shows there is no way the P4 or dual-Xeons are gonna hold a candle to a single Opteron/A64 let alone a dual system. Intel is hurting as their 90 nano stuff is gonna be too little, too late, and the suckers draw a pile of current = a ton of heat. Who wants that in a system thats' too expensive and that under performs??? They should just scrap the Prescott and work on something worth releasing to the market before it's obsolete.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
29 Comments
Back to Article
Kiwi42084 - Tuesday, March 2, 2004 - link
This Benchmark is totally unfair!!!!!The PC-DL has the avalibility to produce 4 usable CPUs...2 Physical and 2 Logical....
Windows XP will only see 2 of these. The benchmarks are being done with Half of the POWER POTENTIAL.
vppaul - Wednesday, February 25, 2004 - link
I am having a problem with my PC-DL board. The System Management BIOS is reporting that there is 4096 MB of RAM, but Windows reports that 2048 MB is available. Anybody ideas? I am trying to get help from Asus tech support but any help would be appreciated. Running dual 3.06 with 4 x 1GB sticks.piperfect - Monday, February 23, 2004 - link
I totally agree with FutureShock999. Why not run several instances of the a divx encoder. For instance do a 2 part movie on an encoder that runs two threads per instance then run both of the parts at the same time in two instances of the program on the xeon and see who finishes first. What you guys are doing is like comparing an f-16 fighter jet to a f-15 and saying the f-15 can only use one of its engines.piperfect - Wednesday, February 11, 2004 - link
I got my pc-dl with 2.8GHz Xeons with raptor drives in RAID0 to 3361MHz 160FSB with a dram clock of 200MHz. 4:5 It ran Sandra burnin overnight. Maybe my board is a newer revision. I read this article after I bought the Raptor so I didn't try to overclock it until yeaterday but it did and it runs well!!!!!FutureShock999 - Monday, October 6, 2003 - link
Wesley,A nice review, but I believe the wrong benchmarks. No one should buy a dual-processor machine to execute a single-threaded application faster, especially when that single proc is slower than others on the market. Dual-proc boards are bought to do multi-threaded stuff, either runnig a single multi-threaded application, or running several different single-threaded applications. In no place did I see benchmarks that explicitly looked like that.
As such, your review was a good cautionary tale for people that didn't KNOW the above, and hopefully will stop some people from spending a lot of money on a this board hoping to have a great UT setup.
Now if you had shown what it was like to play a game, encode media, and download a few gigs of content SIMULTANEOUSLY, then we could really see how this board stacks up to the competition you evaluated it against. And I think it might have beaten them rather handily...
Wesley Fink - Tuesday, September 9, 2003 - link
#23 -We recently changed our standard video card from nVidia's Ti4600 to the ATI RADEON 9800 PRO. Theoretically this should have no impact on encoding scores, but we reran all benchmarks with our new standard hadware on a few of the highest performing boards. Evan ran about half the new encoding benchmarks on the west coast, and I ran the other half on the east coast. As you can see our new results compare very well to each other.
I have no other explanation, but perhaps Evan can shed some light on this. I have used the new ATI Radeon 9800 PRO from day 1 and my benchmarks have been cumulative over the last couple of months with no dramatic change that you point out.
Anonymous User - Tuesday, September 9, 2003 - link
The 865/875 P4 boards in this test all perform 50% faster in the media encoding benchmark in this review than they did in the round-up a couple weeks ago.Has anyone else noticed this dicrepancy? The Abit IS7-G has gone from 64.45 fps to 104 fps. Both times they used a P4 3.0Ghz, 800mhz FSB, with HT enabled. Nothing seems to have changed except the encoding speed. I wish I could do that to my rig ; )
A 50% increase in as linear and consistent a benchmark as DivX encoding is simply astounding.
I just wanted to point this out to everybody around here.
Thank you for your time.
Anonymous User - Tuesday, September 9, 2003 - link
It' will not overclock with the SATA drives on the intel controller. BUT it will if you run them on the onboard promise controller. I have 2 WD Raptor's running RAID0 and the 2800/533's running @3250 100% stable.Anonymous User - Monday, September 8, 2003 - link
Which brings up the question about whether or not HT was enabled on it. And the fact is they used a regular consumer card for video Most workstations would have a workstation class card in them such as a Quadro, FireGL, or even a 3DLabs card in them. It all depends on the applications that one uses. Most normal people wouldn't use a dual machine for gaming anyways. They'd use them for graphical processing or media encoding or file serving and such. Just talk to those guys over at 2CPU.com they know what I'm talking about. ;)Anonymous User - Monday, September 8, 2003 - link
From the looks of it, there would be little if any reason to spend gobs of extra money on a system that is beat by AMD in gaming, and by single P4 siblings in high-end workstation tasks.Anonymous User - Monday, September 8, 2003 - link
Xeon DPs are still at the 533 right now. Xeon MPs are worse off at just 400. Faster Xeons are just around the corner though.sprockkets - Monday, September 8, 2003 - link
Aren't the Xenons at 667FSB? But as usual, they share the bandwidth, which is really bad since they crave bandwidth, but good for memory access. But clearly the HT idea is faster.Running though at 800 maybe really too hard to do, or not ?
Anonymous User - Monday, September 8, 2003 - link
Remember Xeons were built for the server market where stability is king. Intel knew this and this is why the 533MHz FSB has stuck around so long. As Intel rolls out more 800Mhz FSB chipsets and P4 chips the 800Mhz FSB Xeon should follow several months later. The same process could be seen back when the 400Mhz FSB Xeons were around.Anonymous User - Monday, September 8, 2003 - link
Well don't take my word for it, wait for the results yourself and then you can admit you are a fool! Clueless kids looking to argue... What a waste of time.Anonymous User - Sunday, September 7, 2003 - link
Is HyperThreading enabled for those Dual Xeons?Anonymous User - Sunday, September 7, 2003 - link
@13 "it would not improve performance enough to even be competitive"lol... what reviews have you been reading? the extra fsb makes a huge difference! THG overclocked the Xeons by 7 mhz and averaged somewhere around 4-4.5% performance increase! That's pretty damn significant. Even bringing the Xeons up to 667 FSB would make it a very good performer!
Anonymous User - Saturday, September 6, 2003 - link
Don't believe everything you read... ESPECIALLY from Intel. Go read THG's review and *maybe* you'll have a better appreciation for why Intel ain't likely to release a 800 MHS FSB Xeon any time soon and it would not improve performance enough to even be competitive when a 2.0 Gig. Opteron/A64 blows the doors off a dual 3.06 Gig. Xeon. If you wanna feel bad... look at how the dual Opteron beats up on a dual Xeon.Anonymous User - Saturday, September 6, 2003 - link
Read the news -- Intel has said it WILL produce 800Mhz FSB Xeons (when is a bit of a question: the last report I read over a month ago said Q1 '04 but they may have moved it up -- that's what this article seems to imply).Anonymous User - Saturday, September 6, 2003 - link
There are technical reasons why the Xeon can't do 800 MHz FSB. Tom's Hardware went into some of the reasons in their dual Xeon test where Opteron also smoked the Xeon even with the new L3 cache. The reality is that Intels current processors have just become obsolete with Opteron and A64. The benches for A64 will show equal or better performance for a single CPU than Opteron. For multi-processor systems above (4) CPUs the Opteron has an advantage over the A64. As the benches show Intel simply has nothing to compete in either the desktop or server segments and soon the A64 will be available in laptop to so Intel needs to get it in gear. Needless to say all Opteron/A64 CPUs will run both 32-bit and 64-bit O/S's. Linux has been available and Windoze will be available soon. AMD has definitety made a qunatum performance leap over Intel this time, no doubt.Anonymous User - Saturday, September 6, 2003 - link
Remember that server boards have to go through more stringent evaluation then desktop boards. I don't think Intel even considers 800Mhz FSB stable enough for their Xeons yet, and it's a shame. There is only so much increasing your L2 cache size can do for you. I don't think you can use 2.4Cs on this board even if you could find a socket adapter -- remember the 'overclocking settings' were limited to 133->165 (that's almost the 667Mhz FSB that Intel might move the Xeon to soon). So you'd be underclocking it by at least 17%, and at that point why bother? That's if you can find some video card that will take a 79Mhz AGP speed -- good luck with that.Anonymous User - Saturday, September 6, 2003 - link
This review makes, me wonder...is it possible to use two P4 2400/800 CPU's on this board, maybe with some kind of adapter. Those P4's would likely outperform this Xeon setup and possibly also the Opteron, with a very interesting price tag, even if these adaptors would cost $50 each.And if anyone should see an Intel engineer, just tell him we want a 1000 MHZ FSB for those Xeons, as the i875 would surely pass all validation test at this speed, and the PC4000 memory is already available.
Wesley Fink - Saturday, September 6, 2003 - link
The 3200+ Athlon we use for testing AMD IS a Barton core. The 3.0 is the Intel CPU we have been using for benchmarking Intel.Anonymous User - Saturday, September 6, 2003 - link
Why not just make a dual 800FSB P4 system. With HT activated you'd get 4 virtual processors, something that XP Pro can handle.Anonymous User - Saturday, September 6, 2003 - link
i see commens like like its they use thw 3.2Hz p4 but what performance they all expect? 80% from 3.0 to 3.2Ghz?>Lonyo - Saturday, September 6, 2003 - link
"we are reviewing the PC-DL, and comparing it to the performance from the current top Pentium 4, Barton, and Opteron/Athlon64 boards that we have tested"Performance Test Configuration
Processor(s): Intel Dual Xeon 3.06 (1 Mb Cache) 533FSB
AMD Opteron Socket 940 at 2.0GHz (9x222) 444FSB
Intel Pentium 4 at 3.0GHz (800FSB)
No Barton in there, the top Opteron (or so I thought) was the 1.8GHz, and the top P4 is the 3.2 as far as I'm aware.
Anonymous User - Saturday, September 6, 2003 - link
Dual Xeons is a joke. The P4 is starved for bandwidth in a SINGLE processor configuration at 533Mhz FSB. What makes anyone think TWO processors can handle being that starved? It's like eating spaghetti through a straw -- you can get a few noodles, but it's faster to go buy a fork. And, last I heard, Intel wasn't planning on bring a 800Mhz FSB to the Xeon, just a stopgap 667Mhz FSB. Almost any task can be performed better by a single P4 3.0Ghz processor then dual Xeons of any speed on any platform. Xeons make no sense as a consumer platform all. And, with Opteron here, they make no sense as a server platform, either.When can I see my Athlon64 vs. Pentium 4 benchies? I hardly think an overclocked Opteron counts. After all, the 2.0Ghz Athlon64 isn't going to run with dual-channel memory, anyway. Can you say Socket 754? Luckily, AMD is already moving to Socket 940/939 for future releases. Mmm. 64-bit goodness.
Anonymous User - Saturday, September 6, 2003 - link
I'm always dissapointed in dual cpu system reviews, as they never show how the systems multitask. I would like to see the FPS in UT3 while encoding a movie at the same time. Can you play games while burning a DVD? I would like to see the performance while performing multiple tasks at the same time.Anonymous User - Saturday, September 6, 2003 - link
Yea, I'll bet Intel thought they could Spin dual Xeons as a viable filler until Prescott arrives some day. Reality shows there is no way the P4 or dual-Xeons are gonna hold a candle to a single Opteron/A64 let alone a dual system. Intel is hurting as their 90 nano stuff is gonna be too little, too late, and the suckers draw a pile of current = a ton of heat. Who wants that in a system thats' too expensive and that under performs??? They should just scrap the Prescott and work on something worth releasing to the market before it's obsolete.Anonymous User - Saturday, September 6, 2003 - link
Somehow this review looks eerily familiar. Hrm, wonder why :P.