Intel Woodcrest: the Birth of a New King
by Jason Clark & Ross Whitehead on July 13, 2006 12:05 AM EST- Posted in
- IT Computing
Multiple Load Points
For AnandTech Database Benchmarks, we have always focused on "real world" Benchmarks. To achieve this, we have used real applications with loads such that CPU utilization was 80-90%. Recently we discussed how most Enterprise Database Servers do not average 80-90% CPU utilization, but rather something closer to the 30-60% range. We thought it would make more sense to show performance where it is most likely going to be used, as well as the saturation numbers for the situations where the CPU is maxed. We feel this is consistent with how GPUs are reviewed, and how you might test drive a car. With GPUs, the cards are tested with varying resolutions, and anti-aliasing levels. With a car, you don't just hit the highway and see what the top end is.
We settled on six load points for testing. These load points are consistent across all platforms and are throttled from the client, independent of the platform being measured. We chose these load points as they split the load range into 6 roughly equal parts and allow us to extrapolate data between the points. The last/highest load point is a "saturation plus" load point to verify that we tested up the capability of the CPUs.
For any given load point, there is a defined number of threads. Each test is 20 minutes in duration, which includes an 8 minute warm up period followed by a 12 minute measured period. For a given load point, the client submits requests to the DB server as fast as the DB server will respond. The rate which the client is able to submit requests is measured during the final 12 minutes of the test and averaged to determine the Orders/Minute for Dell and Transactions/Minute for Forums. After much blood, sweat, and almost tears we were able to produce repeatable loads with an average deviation of 1.6%.
For each platform we ran the test 5 times for each load point and then averaged the 5 results. This was repeated for all loads, all tests, on all platforms... that is 300 test executions!!! (We won't even get into the debugging issues we had to deal with prior to the final results.) Thankfully, we managed to automate the process as much as possible when implementing the throttling mechanism for the load points.
For AnandTech Database Benchmarks, we have always focused on "real world" Benchmarks. To achieve this, we have used real applications with loads such that CPU utilization was 80-90%. Recently we discussed how most Enterprise Database Servers do not average 80-90% CPU utilization, but rather something closer to the 30-60% range. We thought it would make more sense to show performance where it is most likely going to be used, as well as the saturation numbers for the situations where the CPU is maxed. We feel this is consistent with how GPUs are reviewed, and how you might test drive a car. With GPUs, the cards are tested with varying resolutions, and anti-aliasing levels. With a car, you don't just hit the highway and see what the top end is.
We settled on six load points for testing. These load points are consistent across all platforms and are throttled from the client, independent of the platform being measured. We chose these load points as they split the load range into 6 roughly equal parts and allow us to extrapolate data between the points. The last/highest load point is a "saturation plus" load point to verify that we tested up the capability of the CPUs.
For any given load point, there is a defined number of threads. Each test is 20 minutes in duration, which includes an 8 minute warm up period followed by a 12 minute measured period. For a given load point, the client submits requests to the DB server as fast as the DB server will respond. The rate which the client is able to submit requests is measured during the final 12 minutes of the test and averaged to determine the Orders/Minute for Dell and Transactions/Minute for Forums. After much blood, sweat, and almost tears we were able to produce repeatable loads with an average deviation of 1.6%.
For each platform we ran the test 5 times for each load point and then averaged the 5 results. This was repeated for all loads, all tests, on all platforms... that is 300 test executions!!! (We won't even get into the debugging issues we had to deal with prior to the final results.) Thankfully, we managed to automate the process as much as possible when implementing the throttling mechanism for the load points.
Facebook
Twitter
RSS
59 Comments
View All Comments
Viditor - Thursday, July 13, 2006 - link
Thanks Jason...it's just that the power draw is such a central theme to the review, it would be nice to know exactly what and how things were being tested. Could you let us know what features and chipset were on the OEM Woodcrest system?
Cheers
FesterOZ - Thursday, July 13, 2006 - link
JasonBased on your own Extech tests, the 280 you previously tested here http://www.anandtech.com/showdoc.aspx?i=2644&p...">http://www.anandtech.com/showdoc.aspx?i=2644&p... was drawing a max of 265 watts for that database forum test at max load. Now the same CPU seems to be drawing 300+ Watts.
Can you please explain this variation?
johnsonx - Thursday, July 13, 2006 - link
Between the two platforms, the motherboards (Tyan 2882 vs 2891) are different with different chipsets (AMD+AMD vs NVidia+AMD), the software load is different (I presume even at apparent 100% load, there will be small power consumption differences depending on what the software actually does), and there is probably a power supply difference.Those factors combined could easily account for 40 or so watts.
FesterOZ - Thursday, July 13, 2006 - link
Sorry but I really have to disagree here. 40 Watts in a south bridge? Both tests the CPUS were maxed out.Viditor - Thursday, July 13, 2006 - link
Could you be more specific on the Intel OEM system specs? (we can look up the Tyan of course)I assume you used the same PSU for both systems...
Did you measure power draw at the wall?
MrKaz - Thursday, July 13, 2006 - link
Was this review/preview made by Intel?I mean:
-Charts have blue background (a la Intel). (There is one Anandtech symbol in the charts but could be Intel no difference).
-Charts with performance/watts, since when Anandtech review CPU based on those parameters? Never. Why now?
-.90nm CPU vs .65nm CPU - .65nm CPU wins, higher clock, lower consuming. No big surprise here, especially if this processor is derivates from a mobile one.
When AMD releases one socket F Opteron will be interesting, especially if = Intel with just half the cache size and 3 years "old" design on .90nm process.
fitten - Thursday, July 13, 2006 - link
Heh, by your standards then, the Pentium4s should have been faster than all since they are higher clock.
MrKaz - Thursday, July 13, 2006 - link
Intel.http://www.intel.com/performance/server/xeon/intth...">http://www.intel.com/performance/server/xeon/intth...
sprockkets - Thursday, July 13, 2006 - link
65nm smaller than 90nm, I would hope they consume less power. Still, Intel has always been good with power with the P6 derivatives.berat556 - Thursday, July 13, 2006 - link
I am not impressed with this latest review especially when you take into consideration how the intel is a 3.0GHZ processor while the opteron 285 is 2.6GHZ processor. I was so exited about Conroe but this latest benchmark made put my pre-order order on hold to see if Conroe will do what they said it would a 2.67GHZ win vs. a 2.8GHZ. Major bumper, it looks like woodcrest is still behing the opteron, maybe that is why dell still decided to go with AMD after Intel announced the core architecture.