Perpetual benchmark thread...

[Jeff Gilchrist]

Here is my Q9550 overclocked to 3.2GHz

Processor: Q9550 @ 3.2 GHz (overclocked)
Motherboard: ASUS Maximus II Formula
OS: Vista 64-bit
Memory: 4GB DDR2-1066
mprime: 25.6 build 6 Prime95 Windows64

Timing 53 iterations at 768K FFT length.  Best time: 12.772 ms.
Timing 45 iterations at 896K FFT length.  Best time: 15.534 ms.
Timing 39 iterations at 1024K FFT length.  Best time: 17.695 ms.
Timing 31 iterations at 1280K FFT length.  Best time: 22.183 ms.
Timing 26 iterations at 1536K FFT length.  Best time: 27.177 ms.
Timing 22 iterations at 1792K FFT length.  Best time: 32.319 ms.
Timing 19 iterations at 2048K FFT length.  Best time: 35.918 ms.
Timing 15 iterations at 2560K FFT length.  Best time: 47.232 ms.
Timing 13 iterations at 3072K FFT length.  Best time: 57.917 ms.
Timing 11 iterations at 3584K FFT length.  Best time: 69.005 ms.
Timing 10 iterations at 4096K FFT length.  Best time: 77.161 ms.
Timing 10 iterations at 5120K FFT length.  Best time: 98.245 ms.
Timing 10 iterations at 6144K FFT length.  Best time: 118.966 ms.
Timing 10 iterations at 7168K FFT length.  Best time: 143.937 ms.
Timing 10 iterations at 8192K FFT length.  Best time: 157.418 ms.
Timing FFTs using 2 threads.
Timing 53 iterations at 768K FFT length.  Best time: 6.768 ms.
Timing 45 iterations at 896K FFT length.  Best time: 8.119 ms.
Timing 39 iterations at 1024K FFT length.  Best time: 9.421 ms.
Timing 31 iterations at 1280K FFT length.  Best time: 11.453 ms.
Timing 26 iterations at 1536K FFT length.  Best time: 14.212 ms.
Timing 22 iterations at 1792K FFT length.  Best time: 16.991 ms.
Timing 19 iterations at 2048K FFT length.  Best time: 19.152 ms.
Timing 15 iterations at 2560K FFT length.  Best time: 24.908 ms.
Timing 13 iterations at 3072K FFT length.  Best time: 30.536 ms.
Timing 11 iterations at 3584K FFT length.  Best time: 36.099 ms.
Timing 10 iterations at 4096K FFT length.  Best time: 40.889 ms.
Timing 10 iterations at 5120K FFT length.  Best time: 51.907 ms.
Timing 10 iterations at 6144K FFT length.  Best time: 63.384 ms.
Timing 10 iterations at 7168K FFT length.  Best time: 75.904 ms.
Timing 10 iterations at 8192K FFT length.  Best time: 84.553 ms.
Timing FFTs using 3 threads.
Timing 53 iterations at 768K FFT length.  Best time: 8.935 ms.
Timing 45 iterations at 896K FFT length.  Best time: 10.929 ms.
Timing 39 iterations at 1024K FFT length.  Best time: 17.841 ms.
Timing 31 iterations at 1280K FFT length.  Best time: 10.351 ms.
Timing 26 iterations at 1536K FFT length.  Best time: 12.496 ms.
Timing 22 iterations at 1792K FFT length.  Best time: 14.918 ms.
Timing 19 iterations at 2048K FFT length.  Best time: 16.703 ms.
Timing 15 iterations at 2560K FFT length.  Best time: 22.066 ms.
Timing 13 iterations at 3072K FFT length.  Best time: 26.825 ms.
Timing 11 iterations at 3584K FFT length.  Best time: 31.294 ms.
Timing 10 iterations at 4096K FFT length.  Best time: 35.556 ms.
Timing 10 iterations at 5120K FFT length.  Best time: 43.168 ms.
Timing 10 iterations at 6144K FFT length.  Best time: 51.974 ms.
Timing 10 iterations at 7168K FFT length.  Best time: 62.648 ms.
Timing 10 iterations at 8192K FFT length.  Best time: 69.280 ms.
Timing FFTs using 4 threads.
Timing 53 iterations at 768K FFT length.  Best time: 9.207 ms.
Timing 45 iterations at 896K FFT length.  Best time: 10.059 ms.
Timing 39 iterations at 1024K FFT length.  Best time: 16.015 ms.
Timing 31 iterations at 1280K FFT length.  Best time: 9.511 ms.
Timing 26 iterations at 1536K FFT length.  Best time: 10.154 ms.
Timing 22 iterations at 1792K FFT length.  Best time: 11.980 ms.
Timing 19 iterations at 2048K FFT length.  Best time: 13.849 ms.
Timing 15 iterations at 2560K FFT length.  Best time: 17.726 ms.
Timing 13 iterations at 3072K FFT length.  Best time: 21.741 ms.
Timing 11 iterations at 3584K FFT length.  Best time: 25.498 ms.
Timing 10 iterations at 4096K FFT length.  Best time: 29.050 ms.
Timing 10 iterations at 5120K FFT length.  Best time: 36.026 ms.
Timing 10 iterations at 6144K FFT length.  Best time: 43.255 ms.
Timing 10 iterations at 7168K FFT length.  Best time: 51.299 ms.
Timing 10 iterations at 8192K FFT length.  Best time: 57.794 ms.
Timing trial factoring of M35000011 with 58 bit length factors.  Best time: 2.657 ms.
Timing trial factoring of M35000011 with 59 bit length factors.  Best time: 2.643 ms.
Timing trial factoring of M35000011 with 60 bit length factors.  Best time: 3.021 ms.
Timing trial factoring of M35000011 with 61 bit length factors.  Best time: 3.161 ms.
Timing trial factoring of M35000011 with 62 bit length factors.  Best time: 3.623 ms.
Timing trial factoring of M35000011 with 63 bit length factors.  Best time: 4.645 ms.
Timing trial factoring of M35000011 with 64 bit length factors.  Best time: 5.032 ms.
Timing trial factoring of M35000011 with 65 bit length factors.  Best time: 5.580 ms.
Timing trial factoring of M35000011 with 66 bit length factors.  Best time: 5.551 ms.
Timing trial factoring of M35000011 with 67 bit length factors.  Best time: 5.532 ms.
Benchmark complete.
Work thread stopped.

[fivemack]

Originally Posted by uigrad

I was actually looking into this today also. If you search online for "penryn v8", you'll find lots of articles showing benchmarks from this mythical 8-core processor that would be out by the end of 2007. Apparently, it never did come out, and has been dropped altogether, since I can't find any recent references to it.

'V8' was an early codename for what's now called 'Skulltrail', an Intel marketing plan to convince wealthy gamers to buy two-socket workstations (or, if you prefer, a plan to produce a slightly more affordable dual-Xeon motherboard).

There is a processor called Dunnington out now, which has six Penryn cores on it, but it costs several thousand dollars and only fits in quad-Xeon motherboards which cost several thousand dollars and require four of them.

[xtreme2k]

32bit Prime95 25.6

Intel(R) Core(TM)2 Quad  CPU   Q9450  @ 2.66GHz
CPU speed: 3204.02 MHz, 4 cores
CPU features: RDTSC, CMOV, Prefetch, MMX, SSE, SSE2
L1 cache size: 32 KB
L2 cache size: 6144 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
TLBS: 256
Prime95 32-bit version 25.6, RdtscTiming=1
Best time for 768K FFT length: 12.872 ms.
Best time for 896K FFT length: 15.555 ms.
Best time for 1024K FFT length: 17.786 ms.
Best time for 1280K FFT length: 22.290 ms.
Best time for 1536K FFT length: 27.279 ms.
Best time for 1792K FFT length: 32.463 ms.
Best time for 2048K FFT length: 36.036 ms.
Best time for 2560K FFT length: 47.758 ms.
Best time for 3072K FFT length: 58.197 ms.
Best time for 3584K FFT length: 69.327 ms.
Best time for 4096K FFT length: 77.338 ms.
Best time for 5120K FFT length: 99.106 ms.
Best time for 6144K FFT length: 119.379 ms.
Best time for 7168K FFT length: 144.736 ms.
Best time for 8192K FFT length: 158.121 ms.
Timing FFTs using 2 threads.
Best time for 768K FFT length: 6.939 ms.
Best time for 896K FFT length: 8.319 ms.
Best time for 1024K FFT length: 9.725 ms.
Best time for 1280K FFT length: 11.822 ms.
Best time for 1536K FFT length: 14.530 ms.
Best time for 1792K FFT length: 17.200 ms.
Best time for 2048K FFT length: 19.240 ms.
Best time for 2560K FFT length: 25.383 ms.
Best time for 3072K FFT length: 30.897 ms.
Best time for 3584K FFT length: 36.806 ms.
Best time for 4096K FFT length: 41.227 ms.
Best time for 5120K FFT length: 52.529 ms.
Best time for 6144K FFT length: 63.828 ms.
Best time for 7168K FFT length: 76.876 ms.
Best time for 8192K FFT length: 84.774 ms.
Timing FFTs using 3 threads.
Best time for 768K FFT length: 9.387 ms.
Best time for 896K FFT length: 10.658 ms.
Best time for 1024K FFT length: 17.723 ms.
Best time for 1280K FFT length: 10.713 ms.
Best time for 1536K FFT length: 12.440 ms.
Best time for 1792K FFT length: 14.999 ms.
Best time for 2048K FFT length: 16.796 ms.
Best time for 2560K FFT length: 21.727 ms.
Best time for 3072K FFT length: 26.619 ms.
Best time for 3584K FFT length: 31.049 ms.
Best time for 4096K FFT length: 35.186 ms.
Best time for 5120K FFT length: 43.111 ms.
Best time for 6144K FFT length: 51.869 ms.
Best time for 7168K FFT length: 62.237 ms.
Best time for 8192K FFT length: 69.104 ms.
Timing FFTs using 4 threads.
Best time for 768K FFT length: 8.629 ms.
Best time for 896K FFT length: 9.488 ms.
Best time for 1024K FFT length: 14.996 ms.
Best time for 1280K FFT length: 8.597 ms.
Best time for 1536K FFT length: 10.170 ms.
Best time for 1792K FFT length: 11.987 ms.
Best time for 2048K FFT length: 13.756 ms.
Best time for 2560K FFT length: 17.646 ms.
Best time for 3072K FFT length: 21.556 ms.
Best time for 3584K FFT length: 25.254 ms.
Best time for 4096K FFT length: 28.986 ms.
Best time for 5120K FFT length: 35.781 ms.
Best time for 6144K FFT length: 42.893 ms.
Best time for 7168K FFT length: 50.877 ms.
Best time for 8192K FFT length: 57.381 ms.
Best time for 58 bit trial factors: 3.240 ms.
Best time for 59 bit trial factors: 3.245 ms.
Best time for 60 bit trial factors: 3.231 ms.
Best time for 61 bit trial factors: 3.231 ms.
Best time for 62 bit trial factors: 5.498 ms.
Best time for 63 bit trial factors: 5.508 ms.
Best time for 64 bit trial factors: 5.101 ms.
Best time for 65 bit trial factors: 5.065 ms.
Best time for 66 bit trial factors: 5.068 ms.
Best time for 67 bit trial factors: 5.057 ms.

[petrw1]

Processor: Intel Core 2 Quad Q9450 (2.66 Ghz, 12Mb Cache, 1333 FSB)
Motherboard: Intel Desktop DG33BU/USB 2.0 Dual Channel DDR/1333/1066/800
Memory: 4GB DDR2 PC2-6400 (800Mhz)

===================================================

Processor: Intel Core 2 Extreme QX9650 Quad-Croe (3.00 Ghz, 12Mb Cache, 1333 FSB, 64 Bit EMT 64, Intel Virtualization, 45 nm)
Motherboard: P5K - P35, Dual DDR2-1066, Quad-Core and Penryn Support, 1333 Mhz, IDE+SATA II + RAID + eSATA
Memory: 2GB PC3-8500/DDR1066 DDR3 Kingston

===================================================

These are from two different suppliers. I tried to make everything else equal and see how much more I would pay going from Q9450 to Q9650 and from DDR2 800 to DDR3 1066. Can someone tell me if the Motherboards are essentially equivalent or NOT. System 2 is about $800 more ($1,200 vs $2000). From the quote it has half the memory but if I buy system 2 I will up it to 4Gb ... unless someone tells me it is a waste.

If I read the benchmarks page correctly system 2 should be about 12% faster for current LL tests. Does this sound right?

[Cruelty]

If I read it right both motherboards support DDR2 memory... from the two mentioned I would go for the second option although ASUS is not my favourite motherboard brand. Also I would buy 2x2GB at the beginning because upgrading dual-channel configuration later on is tricky (used modules should be generally the same).
Both motherboards are not equal (different chipsets), Intel uses chipset with integrated GFX (which will have an impact on performance). Asus motherboard officialy supports DDR2-1066 memory while Intel only DDR2-800. Go for the second option although check if you can exchange Asus for some Gigabyte motherboard, e.g. EP35-DS3 or DS4 series

[James Heinrich]

I'm not sure if this belongs in the benchmark thread, but I'd say forget both the Intel board (I've never liked them) and the QX-series CPU (waste of money better spent elsewhere). I agree with Cruelty on recommending Gigabyte boards (but Asus is good too). The P5K series is all DDR2 (except for the P5K-C which accepts either DDR2 or DDR3). If you go DDR2, get 1066MHz. If you get DDR3, go for 1600MHz. Inquire about the Q9550, it should cost almost identical to the Q9450, but at 2.83GHz (instead of 2.66GHz).

[petrw1]

Thanks for the comments ...

I talked to a local veteran supplier about motherboards:
In his opinion, based on service repair frequency of Motherboards specifically:
"Intel is crap. Gigabyte comes in for warranty work a lot more often than ASUS does"

Based on your feedback I will go for the Q9550 instead of Q9450 and, if it is not a lot more pricey, I will get 2X2GB DDR3 1066 RAM.

[James Heinrich]

Originally Posted by petrw1

"Intel is crap. Gigabyte comes in for warranty work a lot more often than ASUS does"

I'd dispute that opinion based on what I see at work, but Asus does outsell Gigabyte at least 4:1. (Personally, I've bought only Asus boards from 1996-2006, but my last 2 systems have been Gigabyte. But honestly I'd be happy with either brand. Except for Asus' ROG series of boards (Striker, Rampage, etc), way too many problems for me to trust those with anything, but that's another story).

Originally Posted by petrw1

I will get 2X2GB DDR3 1066 RAM.

You don't want to do that. If you do DDR2, go 1066MHz. If you do DDR3, do 1600MHz (or faster, with caution). DDR3 @ 1066MHz will be slower than DDR2 @ 1066MHz, you need a noticeable clockspeed advantage to get more speed out of DDR3 than DDR2.

[petrw1]

Originally Posted by James Heinrich

You don't want to do that. If you do DDR2, go 1066MHz. If you do DDR3, do 1600MHz (or faster, with caution). DDR3 @ 1066MHz will be slower than DDR2 @ 1066MHz, you need a noticeable clockspeed advantage to get more speed out of DDR3 than DDR2.

Why would OLDER DDR2 be faster than DDR3? (sorry if this sounds like an uneducated question).

Not sure I can afford DDR3 at 1600 Mhz, but I also was warned against the 3 year old DDR2 technology.

Please define "noticeable clockspeed advantage". A Q9550 at 2.83 Mhz is pretty fast nowadays.

[sylvester]

Originally Posted by petrw1

Why would OLDER DDR2 be faster than DDR3?

Lower latency, e.g. 5-5-5-15 (DDR2) vs. 7-7-7-15 (DDR3).

See Wikipedia article about DDR3 for starters.

If you need further explanation, google "bandwidth latency tradeoff". There are plenty of articles about this general subject, so you may want to narrow it down to your favourite site, e.g. tomshardware.com.

[db597]

Originally Posted by petrw1

Please define "noticeable clockspeed advantage". A Q9550 at 2.83 Mhz is pretty fast nowadays.

James means you need DDR3 at 1600MHz to beat DDR2 at 1066MHz. It's the memory clockspeed and not the CPU clockspeed. If they are the same clockspeed, the DDR2 will be faster due to lower latency, as Sylvester has pointed out.

We pretty much had the same issue when moving from DDR to DDR2 as well. I remember latencies for DDR was around 3. Now we're looking at DDR2 at 5, and DDR3 at 7 (or even 9).