Author: William Henning
Editor: Howard Ha
Publish Date: Wednesday, September 12th, 2007
Originally Published on Neoseeker (http://www.neoseeker.com)
Article Link: http://www.neoseeker.com/Articles/Hardware/Reviews/x3210_overclock/
Copyright Neo Era Media, Inc. - please do not redistribute or use for commercial purposes.
Now that dual core processors have become main stream, the next push is for quad core processors.
Last November, we covered the launch of Intel's QX6700 processor - and found that while performance was excellent, there were few applications that could take advantage of the additional cores. Time marches on, and some software is beginning to take better advantage of multiple cores.
Recently, we reviewed a Core 2 Duo E6750 that was based on the new G0 stepping of the Core 2 die, and we found it to be by far the best overclocking Core 2 processor to date. When we became aware of how inexpensive the new X3210 Xeons are, most of which are also based on the G0 stepping of the Core 2 die, we had to take one for a spin... after all, it might set new in-house records for quad core overclocks!
The X3210 has two dual core Core 2 Duo dies connected inside the packaging; and it runs at 2.13GHz on a 1066MHz FSB - meaning that it has a maximum 8X multiplier with a 266MHz QDR FSB.
We could tell from the packaging that we had one of the desirable G0 stepping based X3210's; and given our great success overclocking a dual core (single die) E6750, I figured we would be able to get some really nice results.
I was right :-)
Hardware for testing of the Intel X3210 Quad Core Xeon processor:
Software used during testing consisted of the following:
Please note that we are showing overclocked results in all the charts - we are not holding you in suspense until the end of the article. :-) We are also presenting you with a FULL set of overclocked benchmarks - our test sytems were all stable at the settings shown with air cooling, thanks to our Noctua NH-U 12 heatsink and accompanied Noctua 120mm fans.
Please take added note that the chart labels incorporate a lot of information about the test configuration. The first line shows the socket type and the model of the processor. The second line shows the "FSB/HT clock rate" x "CPU multiplier" followed by the effective DDR2 memory speed. Please note further that all of the DDR2 tests were run at 4-4-4-12/2T timings unless otherwise specified.
At stock speeds, the Xeon X3210 had the lowest result here; however it performed quite well when overclocked.
Two things really jump out from these results:
As we would expect, pure clock frequency and chipset efficiency cause the Xeon not to place first here.
Well, what can I say?
Sandra Memory Bandwidth
The Xeon X3210 does very well in the Sandra memory bandwidth benchmark, both at stock and overclocked. When overclocked it is only beaten by the much higher FSB overclocked dual core G0 stepping E6750.
The single threaded WinRAR is significantly affected by three factors: processor speed, FSB speed and memory latency. The X3210 does well here when not running at its (low) stock speed.
The multi-threaded WinRAR benchmark LOVES the X3210 (and QX6700) - it takes good advantage of multiple cores, high FSB and memory bandwidth. The overclocked Xeon DOMINATES this chart!
The Xeon X3210 does well here - especially when overclocked - but it cannot overcome the advantage of the 975X chipset. I can hardly wait to benchmark some of the upcoming X38 (975X's successor) motherboards!
Decent write performance, but the higher FSB E6750 overclock won.
The Xeon X3210 does well here, but it is limited by the P35 chipset - the QX6700 on the 975X based board had far superior latency numbers.
For RightMark bandwidth, the X3210 did extremely well - both at stock and overclocked; matter of fact, it took top spot on the chart!
The Xeon X3210 did well for single core LAME results, especially when overclocked - but the higher OC'd E6750 ofcourse took top spot.
Surprisingly enough, the multi-threaded LAME benchmark was also won by the dual core chip... pointing out a limitation of the benchmark - it cannot make use of more than two cores.
The X3210 did VERY well with TMPGEnc, however TMPGEnc is also obviously limited to two threads.
CineBench LOVES multi-core processors. The more cores, the better :)
The Xeon X3210 does very well at stock, and dominates the chart when overclocked.
As we would expect, POV-RAY, when running the multi-threaded version, can also make excellent use of additional cores; so it should be no surprise that the X3210 does well at stock, and smokes when overclocked!
Call of Duty
The Xeon X3210 starts out not so great at stock - due to its low stock clock speed - however once overclocked, the performance is excellent, only beaten by the higher clocked E6750.
Just like in Call of Duty; at stock speeds the gaming performance is decent, but it really gets going when overclocked.
The performance of the Xeon X3210 is quite decent even at stock, but it really comes alive overclocked. As Doom does not effectively use multiple cores, it is not surprising that the higher clocked E6750 beat it when overclocked; and the overclocked QX6700 was no doubt helped by the excellent 975X chipset.
The results from Halo match the Doom 3 results pretty closely as far as how well the X3210 does relative to the QX6700 and the E6750.
The stock figures for the Xeon X3210 are less than impressive, however when overclocked, they are quite acceptable - especially given that Jedi Knight really only uses one core.
Same story; excellent performance when overclocked - but beaten by a higher clocked dual core.
The Xeon X3210 is the best Core 2 Quad for overclocking that I've had the pleasure to work with to date; no doubt largely due to the G0 stepping used for its dies. The maximum stable processor speed we reached during testing was 3.52GHz – for four cores with air cooling!
In order to reach 3.52GHz, we took the following simple steps:
I was very pleased to see the chip run through all of our test suite with 100% stability at this clock speed.
I was able to get into Windows at up to 3.64GHz, however the system was not 100% stable, so I will only claim the stable 3.52GHz overclock.
The stock power consumption is quite good; only 101W drawn by the whole system (excluding monitor) when idling at the Windows desktop; rising to 145W under maximum load.
When overclocked, the power consumption increases greatly - 116W when idling at 3.52GHz, and a whopping 207W under maximum load at 3.52GHz!
By comparison, at stock the E6750 drew 100W at idle and 132W when loaded, and when overclocked to 3.8GHz it drew 107W idling and 157W at full load.
We'd like to thank NCIX.com for helping us get a Xeon X3210 for this review.
When I reviewed the Core 2 Duo E6750, I was blown away by its air cooled overclocking potential. Ever since then, I have been looking forward to reviewing a Core 2 Quad processor based on the same G0 stepping die - and when the opportunity presented itself for us to get a Xeon X3210 based on that core in, I jumped on it.
I was not dissapointed.
I don't know about you, but to me, a 65.25% overclock of a quad core processor - and only air cooled at that - is very impressive. Granted, in a couple of the benchmarks the QX6700 running at 3.45GHz beat the X3210 running at 3.52GHz - but only because of the superior memory controller in the 975X chipset used for the QX6700 review.
If you do a lot of rendering of 3D images, or MPEG conversions then you currently cannot get a faster processor than a Core 2 based Intel quad core processor. For such highly multi-threaded software, the X3210 will trounce any dual core processor.
Gamers currently would still be slightly better off with a highly overclocked dual core processor as most games don't even take advantage of two cores, never mind four - however as time goes by, and highly multi-threaded games become prevelant, quad core processors will start to trounce dual core processors... and besides which... even running at a "conservative" 3.2GHz overclock at stock Vcore settings will get you more gaming goodness than ANY stock processor can offer!
Please do not redistribute or use this article in whole, or in part, for commercial purposes.