I swear this is not to inject some IBM marketing into this valuable discussion (I have technical role myself within IBM and have very little to do with marketing) but I was wondering if you have evaluated running the same test on a system with a NON-vanilla Intel 8500 chipset. IBM develops its own X3 chipset to power the XEON MP servers in the x366 and x460 servers and that is supposed (actually known) to be much better than the Intel base kit.
is mostly true. One thing that has come out from an AMD consurtium is the Hypertransport v3. This will definitely affect the multi-processor server issue. From clusters to SMP.
Would it be possible to provide details on memory configuration of both systems?
Performance may depend significantly on the specific configuration. So, the test results without this type of detail don't make much sense.
Important details for Intel system:
1) Number of memory boards (this is equivalent to memory channels used)
2) Number of DIMMs per memory board
3) Memory mode (Max Performance, Max Compatibility, RAID, Mirroring)
4) Agjacent Cache Line Prefetch (enabled or disabled)
5) Hardware Prefetcher (enabled or disabled)
Important details for AMD system:
1) Memory speed (DDR400, DDR333, or DDR266)
2) ECC disabled or enabled
3) Chip Kill disabled or enabled
"AMD is reported to have 81.5% of the US retail PC market with Intel sitting at 18.5%". A very interestng statistic. How much bigger would the margin be if the World wide leader in PC sales "Dell" would crack open the door for AMD? Alas though, that will not happen as long as Intel keeps Dell executives and board members pockets padded.
quote: How much bigger would the margin be if the World wide leader in PC sales "Dell" would crack open the door for AMD?
US Retail does not include Dell's direct sales, thus no inherant change in US retail sales figures.
It's a part of a part of a part of the total world-wide microprocessor sales.
Inte have trouble sharing FSB amongst more points. It's easier when it's on CPU but it's not for free. When they first went to dual-core they used 800MHz FSB instead of the EE 1066MHz and was first available for only single socket systems. This pattern continues as Intel introduces new products with limited FSB speed.
Intel does not seem to care on MP side. Only Tulsa (last release in netburst?) with insane amount of cache is releasing. They will not be competitive until Tigerton releases mid-2007. I guess cancellation of whitefield is the culprit.
"AMD is completely quiet about anything other than a socket and memory controller change;"
Yes we heard that new Opteron 1xx would be on AM2 and Opteron 2xx and 8xx would be on the new socket F.
We hear plenty about the AM2 launch (including being brought forward, and motherboards).
Things have gone quiet on the socket F. What is the latest news Anandtech? Because if everything is moving to DDR2 we ought to buy Opterons AFTER the migration rather than before. And what sort of motherboards are going to be launched to support it? Hoping to see some interesting offerings from the usual suspects like Tyan, Iwill etc.
It just would be nice to hear the odd Anandtech story to confirm things are still on some kind of schedule. If it were not for the upcoming migration I would have bought a cluster of Opterons already.
Coming from a long time enterprise xeon user, its sad that intc can't even keep up.. its even sadder that my infrastructure is already setup and I can't get one of these AMD machines in till we hit our load on the current xeons.. :(
Honestly it doesn't hurt to bring the point home again and again about Opteron. Those making buying decisions who are still stuck on Intel only solutions need the truth laid out for them regularly and repeatedly, imo. Thank you for doing that, because the current Intel solutions are a waste of money, and I hate that the most of all.
Are your test result comments correct? ie DVD Store Test, should the AMD lead be 40% not 29% ?
AMD has 8853 more orders, which is 40% more than 21782.The same error (?) goes for the other percentages mentioned.
If I am wrong please forgive somebody with an old brain !
Right, using funky-maths, if a 2.6GHz Conroe is 20% faster than a 2.8GHz X2.
And this 8-core 2.6GHz (?) Opteron system is 40% faster than a 3GHz 8-core Xeon.
And the issue is scalability, i.e., the same problem will exist for Woodcrest.
However Woodcrest will be 3GHz, but this will only be useful for cache-bound tasks because of the scalability problems (expect all the Intel biased websites to only benchmark Woodcrest with this type of task). Maybe Opteron will be dual-core 3GHz by then...
Anyway I can't conclude anything without making huge vast leaps of guesswork, but yes, it will be very close later this year, but if Woodcrest can't beat Opteron because of the platform limitations then there will be little to no incentive to switch back from buying Opteron based servers to Intel based servers (for the companies that have switched to AMD), apart from power consumption.
However it means Intel will stop losing customers to AMD in the server space, if the performance and power consumption issues are addressed. By no means do I see AMD losing much of their gained marketshare however (and by Q4 it will be higher than now I'm sure).
If power consumption is your concern, then AMD offer reduced power versions of Opteron chips (eg the HE models) at increased cost. When they can migrate to 65nm process, we should see even better power economy.
Here's what I think will happen. AMD will get a run for its money on the single-socket and dual-socket arenas (since dual-FSB pretty much keeps the FSB from being too limiting) but once you scale higher AMD will once again rule the roost. I also think that quad-core will be better on AMD's architecture, too, since throwing 4 Conroe/Woodcrest cores on a single FSB (266MHz I'd expect, though I suppose it could be 333) will give these cores a huge performance hit, more so in MP configurations (just thinking about the cache-coherency traffic for 8 cores scares me)
Maybe, but it's hard to say at this point, how low FSB speed will affect Core Architecture, plus this is the Xeon MP space, where it won't be transistioning to Core Architecture for sometime until Q1 2007 earliest from what I believe with the Clovertown MP/Tigerton Core, I don't think there is a Woodcrest equivalent for this space.
I expect AMD to have the "FSB" advantage in 4-8 way situations yes, however this may or may not translate to a performance crown for AMD.
Thoug I think it's pretty safe for Intel that they can take back the Xeon DP space with Woodcrest, and DIB.
I think you need to remember a couple of things. First, Woodcrest is based off of the next-generation Core architecture. Power and performance characteristics will be completely different from the current NetBurst chips. Second, Woodcrest will also have a 1333 MHz front side bus -- twice as fast as the server benchmarked here. Finally, and I could be wrong on this, but I think Intel is looking at one shared front side bus for every two sockets, and they could potentially move to one front side bus for each socket.
That last idea would make sense for quad-core. And speaking of quad core, that's such a significant change that again it's almost impossible to predict at this stage. I mean, we don't even really know how AMD or Intel are going to build their quad core packages. Are we talking a single die, or will they have multiple dice on a package? Will Intel change the way the dice communicate with each other?
Of course, none of this things means that Intel will come out on top, but there are enough significant changes that we can't declare a victor at this point. Also, large L3 caches can indeed help server work. Otherwise, why would IBM even make a POWER5 chip with 128 MB of L3 cache? Adding tons of cache to a desktop system rarely helps, but enterprise servers are completely different beast.
At the very least, things should get interesting later this year. :-)
In the DVD store test the AMD system is indeed 40% faster than the Intel system, but the Intel system is 30% slower than the AMD one. He seems to have chosen the AMD scores as the reference but incorrectly says that AMD is 30% faster instead of saying that Intel is 30% slower.
What about power consumption and heat dissipation? More heat would require more power for air conditioners? More power to the server room requires, the more money you have to spend to maintain the solution.
quote: We wanted to do power consumption numbers, however the Opteron was a 110V system and the Paxville is a 220V system :)...
What difference does that make exactly? Wattage is wattage. The supply voltage has no effect on total system power draw. Perhaps the Intel box draws too much power for the measuring equipment to handle. ;>)
It MAKES adifference of around 3-5% in favor of the 220V system(whatever system it is)
The same efficiency PSU is generally 3-5% more efficient than 110V one. To convert from 220 to 12V is simply "easier" than to go from 110V. This is also the reason 12V rail is employed for powering CPU's and GPU's PWMs. It is simply more efficient.
However the power comparison would not hurt since Dempsey would be used in real system so the Intel system handicap would be offset somehow.
Seems K8L is gona come right on time for Core MP chips...
Agreed, doesn't make a difference since a co-lo is giving you an certain Amp circuit with your rack, regardless of the voltage you need, and will charge you more per Amp required.
Didnt paxville come out in Q4 last year, why this review now? We all know Paxville was just a stop-the-bleed solution by Intel, to get a Dual-Core Xeon. It was never expected to be a performance part at all. So why this no-brainer review? The market's moved on. A point in Paxville's favour is its virtualization support though.
Even worse, Paxville is an older chip than Dempsey. I have to wonder if Intel's best MP offering is a 3 ghz Paxville chip. Is this true? If so, why? Intel has released 3.6 ghz Dempsey-based Xeons have they not? Dempsey and Paxville aren't too far off from one another, but if I recall correctly, Paxville was based off Smithfield while Dempsey was based off Presler (making Dempsey superior to Paxville).
If Intel can't ship anything better than a 3 ghz Paxville in 4-way configurations, it's no wonder that they're losing. AMD has also released the Opteron 885, meaning the benchmarks we just saw were not run using AMD's best 4-way/8-way chip!
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.
35 Comments
Back to Article
massimor - Friday, April 28, 2006 - link
I swear this is not to inject some IBM marketing into this valuable discussion (I have technical role myself within IBM and have very little to do with marketing) but I was wondering if you have evaluated running the same test on a system with a NON-vanilla Intel 8500 chipset. IBM develops its own X3 chipset to power the XEON MP servers in the x366 and x460 servers and that is supposed (actually known) to be much better than the Intel base kit.More info here:
http://www.realworldtech.com/page.cfm?ArticleID=RW...">http://www.realworldtech.com/page.cfm?ArticleID=RW...
ftp://ftp.software.ibm.com/eserver/benchmarks/wp_X...">ftp://ftp.software.ibm.com/eserver/benchmarks/wp_X...
It would have been interesting to see how the AMD based system would compare to an X3 based system. On other benchmark it's usually a head-to-head.
Massimo.
bjbrock - Wednesday, April 26, 2006 - link
is mostly true. One thing that has come out from an AMD consurtium is the Hypertransport v3. This will definitely affect the multi-processor server issue. From clusters to SMP.Art - Tuesday, April 25, 2006 - link
Would it be possible to provide details on memory configuration of both systems?Performance may depend significantly on the specific configuration. So, the test results without this type of detail don't make much sense.
Important details for Intel system:
1) Number of memory boards (this is equivalent to memory channels used)
2) Number of DIMMs per memory board
3) Memory mode (Max Performance, Max Compatibility, RAID, Mirroring)
4) Agjacent Cache Line Prefetch (enabled or disabled)
5) Hardware Prefetcher (enabled or disabled)
Important details for AMD system:
1) Memory speed (DDR400, DDR333, or DDR266)
2) ECC disabled or enabled
3) Chip Kill disabled or enabled
Thanks,
Artem
johnsonx - Tuesday, April 25, 2006 - link
Also, is the memory on the Opteron running node-interleaved or NUMA? Any comparitive benchmark numbers between the two modes?deathwalker - Tuesday, April 25, 2006 - link
"AMD is reported to have 81.5% of the US retail PC market with Intel sitting at 18.5%". A very interestng statistic. How much bigger would the margin be if the World wide leader in PC sales "Dell" would crack open the door for AMD? Alas though, that will not happen as long as Intel keeps Dell executives and board members pockets padded.tygrus - Tuesday, April 25, 2006 - link
You said,US Retail does not include Dell's direct sales, thus no inherant change in US retail sales figures.
It's a part of a part of a part of the total world-wide microprocessor sales.
Inte have trouble sharing FSB amongst more points. It's easier when it's on CPU but it's not for free. When they first went to dual-core they used 800MHz FSB instead of the EE 1066MHz and was first available for only single socket systems. This pattern continues as Intel introduces new products with limited FSB speed.
trivik12 - Monday, April 24, 2006 - link
Intel does not seem to care on MP side. Only Tulsa (last release in netburst?) with insane amount of cache is releasing. They will not be competitive until Tigerton releases mid-2007. I guess cancellation of whitefield is the culprit.peternelson - Monday, April 24, 2006 - link
"AMD is completely quiet about anything other than a socket and memory controller change;"
Yes we heard that new Opteron 1xx would be on AM2 and Opteron 2xx and 8xx would be on the new socket F.
We hear plenty about the AM2 launch (including being brought forward, and motherboards).
Things have gone quiet on the socket F. What is the latest news Anandtech? Because if everything is moving to DDR2 we ought to buy Opterons AFTER the migration rather than before. And what sort of motherboards are going to be launched to support it? Hoping to see some interesting offerings from the usual suspects like Tyan, Iwill etc.
It just would be nice to hear the odd Anandtech story to confirm things are still on some kind of schedule. If it were not for the upcoming migration I would have bought a cluster of Opterons already.
themelon - Monday, April 24, 2006 - link
Did the DL585 you guys used run the memory at 266MHz? The one that I have does but it is about 2 years old so they may have changed the specs on it.GrammatonJP - Monday, April 24, 2006 - link
Coming from a long time enterprise xeon user, its sad that intc can't even keep up.. its even sadder that my infrastructure is already setup and I can't get one of these AMD machines in till we hit our load on the current xeons.. :(coldpower27 - Monday, April 24, 2006 - link
At least then by the time you do replace your machines hopefully Intel will be more competitive in this spaceAnemone - Monday, April 24, 2006 - link
Honestly it doesn't hurt to bring the point home again and again about Opteron. Those making buying decisions who are still stuck on Intel only solutions need the truth laid out for them regularly and repeatedly, imo. Thank you for doing that, because the current Intel solutions are a waste of money, and I hate that the most of all.:)
johnzo - Monday, April 24, 2006 - link
Are your test result comments correct? ie DVD Store Test, should the AMD lead be 40% not 29% ?AMD has 8853 more orders, which is 40% more than 21782.The same error (?) goes for the other percentages mentioned.
If I am wrong please forgive somebody with an old brain !
Jason Clark - Monday, April 24, 2006 - link
You are correct, my mistake. This is fixed.psychobriggsy - Monday, April 24, 2006 - link
Right, using funky-maths, if a 2.6GHz Conroe is 20% faster than a 2.8GHz X2.And this 8-core 2.6GHz (?) Opteron system is 40% faster than a 3GHz 8-core Xeon.
And the issue is scalability, i.e., the same problem will exist for Woodcrest.
However Woodcrest will be 3GHz, but this will only be useful for cache-bound tasks because of the scalability problems (expect all the Intel biased websites to only benchmark Woodcrest with this type of task). Maybe Opteron will be dual-core 3GHz by then...
Anyway I can't conclude anything without making huge vast leaps of guesswork, but yes, it will be very close later this year, but if Woodcrest can't beat Opteron because of the platform limitations then there will be little to no incentive to switch back from buying Opteron based servers to Intel based servers (for the companies that have switched to AMD), apart from power consumption.
However it means Intel will stop losing customers to AMD in the server space, if the performance and power consumption issues are addressed. By no means do I see AMD losing much of their gained marketshare however (and by Q4 it will be higher than now I'm sure).
peternelson - Monday, April 24, 2006 - link
If power consumption is your concern, then AMD offer reduced power versions of Opteron chips (eg the HE models) at increased cost. When they can migrate to 65nm process, we should see even better power economy.
Furen - Monday, April 24, 2006 - link
Here's what I think will happen. AMD will get a run for its money on the single-socket and dual-socket arenas (since dual-FSB pretty much keeps the FSB from being too limiting) but once you scale higher AMD will once again rule the roost. I also think that quad-core will be better on AMD's architecture, too, since throwing 4 Conroe/Woodcrest cores on a single FSB (266MHz I'd expect, though I suppose it could be 333) will give these cores a huge performance hit, more so in MP configurations (just thinking about the cache-coherency traffic for 8 cores scares me)coldpower27 - Monday, April 24, 2006 - link
Maybe, but it's hard to say at this point, how low FSB speed will affect Core Architecture, plus this is the Xeon MP space, where it won't be transistioning to Core Architecture for sometime until Q1 2007 earliest from what I believe with the Clovertown MP/Tigerton Core, I don't think there is a Woodcrest equivalent for this space.I expect AMD to have the "FSB" advantage in 4-8 way situations yes, however this may or may not translate to a performance crown for AMD.
Thoug I think it's pretty safe for Intel that they can take back the Xeon DP space with Woodcrest, and DIB.
JarredWalton - Monday, April 24, 2006 - link
I think you need to remember a couple of things. First, Woodcrest is based off of the next-generation Core architecture. Power and performance characteristics will be completely different from the current NetBurst chips. Second, Woodcrest will also have a 1333 MHz front side bus -- twice as fast as the server benchmarked here. Finally, and I could be wrong on this, but I think Intel is looking at one shared front side bus for every two sockets, and they could potentially move to one front side bus for each socket.That last idea would make sense for quad-core. And speaking of quad core, that's such a significant change that again it's almost impossible to predict at this stage. I mean, we don't even really know how AMD or Intel are going to build their quad core packages. Are we talking a single die, or will they have multiple dice on a package? Will Intel change the way the dice communicate with each other?
Of course, none of this things means that Intel will come out on top, but there are enough significant changes that we can't declare a victor at this point. Also, large L3 caches can indeed help server work. Otherwise, why would IBM even make a POWER5 chip with 128 MB of L3 cache? Adding tons of cache to a desktop system rarely helps, but enterprise servers are completely different beast.
At the very least, things should get interesting later this year. :-)
Furen - Monday, April 24, 2006 - link
You are indeed correct.In the DVD store test the AMD system is indeed 40% faster than the Intel system, but the Intel system is 30% slower than the AMD one. He seems to have chosen the AMD scores as the reference but incorrectly says that AMD is 30% faster instead of saying that Intel is 30% slower.
Ganjalf - Monday, April 24, 2006 - link
You're correct! The Opteron has a 36-51% lead over the Paxville.Theunis - Monday, April 24, 2006 - link
What about power consumption and heat dissipation? More heat would require more power for air conditioners? More power to the server room requires, the more money you have to spend to maintain the solution.Jason Clark - Tuesday, April 25, 2006 - link
How about a separate article, short but go through powernow and the numbers?Cheers
Jason Clark - Monday, April 24, 2006 - link
We wanted to do power consumption numbers, however the Opteron was a 110V system and the Paxville is a 220V system :)...Cheers
Ecmaster76 - Monday, April 24, 2006 - link
Err, you can measure power on a 220V system too. Its not that hard and the numbers will certainly be comparable.AnandThenMan - Monday, April 24, 2006 - link
What difference does that make exactly? Wattage is wattage. The supply voltage has no effect on total system power draw. Perhaps the Intel box draws too much power for the measuring equipment to handle. ;>)
mino - Monday, April 24, 2006 - link
It MAKES adifference of around 3-5% in favor of the 220V system(whatever system it is)The same efficiency PSU is generally 3-5% more efficient than 110V one. To convert from 220 to 12V is simply "easier" than to go from 110V. This is also the reason 12V rail is employed for powering CPU's and GPU's PWMs. It is simply more efficient.
However the power comparison would not hurt since Dempsey would be used in real system so the Intel system handicap would be offset somehow.
Seems K8L is gona come right on time for Core MP chips...
Lifted - Monday, April 24, 2006 - link
Agreed, doesn't make a difference since a co-lo is giving you an certain Amp circuit with your rack, regardless of the voltage you need, and will charge you more per Amp required.xtremejack - Monday, April 24, 2006 - link
Didnt paxville come out in Q4 last year, why this review now? We all know Paxville was just a stop-the-bleed solution by Intel, to get a Dual-Core Xeon. It was never expected to be a performance part at all. So why this no-brainer review? The market's moved on. A point in Paxville's favour is its virtualization support though.DrMrLordX - Monday, April 24, 2006 - link
Even worse, Paxville is an older chip than Dempsey. I have to wonder if Intel's best MP offering is a 3 ghz Paxville chip. Is this true? If so, why? Intel has released 3.6 ghz Dempsey-based Xeons have they not? Dempsey and Paxville aren't too far off from one another, but if I recall correctly, Paxville was based off Smithfield while Dempsey was based off Presler (making Dempsey superior to Paxville).If Intel can't ship anything better than a 3 ghz Paxville in 4-way configurations, it's no wonder that they're losing. AMD has also released the Opteron 885, meaning the benchmarks we just saw were not run using AMD's best 4-way/8-way chip!
coldpower27 - Monday, April 24, 2006 - link
Xeon MP will take some time to transistion to Core Architecture, were looking at Q1 2007 at the earliest, with the introduction of Tigerton hopefully.In the meantime Intel will update this platform with Tulsa, which still wouldn't be that competitive, but a step in the right direction.
Jason Clark - Monday, April 24, 2006 - link
I agree, it is getting a bit old discussing the older platforms, but this is a fairly new platform. Things will get more interesting later this year.Furen - Monday, April 24, 2006 - link
The 2.8GHz Paxville DP is the chip that came out on Q4, 2005Ozenmacher - Monday, April 24, 2006 - link
I think the title should be Eight Core Servers ;)JarredWalton - Monday, April 24, 2006 - link
Yup. Fixed. My error, not Jason's.