:welcome:I curious how the new Nehalems are doing compared to other multicore/Dual processor systems :p:
Asus z7s with 2x Xeon L5410@2,8Ghz, Vista 64, 4x2Gb FBDimm
Printable View
:welcome:I curious how the new Nehalems are doing compared to other multicore/Dual processor systems :p:
Asus z7s with 2x Xeon L5410@2,8Ghz, Vista 64, 4x2Gb FBDimm
P5Q3 Deluxe+e8600+2x1gb ddr3-1600+4870x2
http://img227.imageshack.us/img227/7...ot33rx9.th.jpg
Intel skulltrail platform @4400 aircooled.
2x 9775 exreme edition
2x 2gb kingston fbdimm
geforce 9800gx2
http://images.hardware.info/usersys/...s/005732-7.jpg
http://www.isarapix.org/pix79/1227967912.png
Nehalem UP, still at stock.
Single- and multicore scores, and multicore speedup are completely off tangent due to the Nehalem quirks.. One, the single core runs at a higher frequency than all the cores do due to Turbo mode. Two, Hyper-Threading. So Hyper-Threading is actually yielding more than 0.11x higher thread parallelism, it's just that the one core is clocked up when it works alone so that number is useless. :p:
http://bildr.no/thumb/312362.jpeg
i didnt know if i was supposed to post the stock settings, but here is my 24/7 settings
and i have no idea if the score is good or bad..lol
here's my QX9650 and 4.4Ghz vs i7 920 at 3.7Ghz no turbo
just a qwick test whit a singel qx9775.
need a beter psu to use the second qx9775.
all @ stock, mem ar 4 x 1 gb kingstone @ 319,2 mhz, 5:4 , 5-5-5-15-20
how do you test with cinebench? do you just push start all tests? this is what I get and it looks weird. My setup is
CPU: intel e6750 2.66ghz
Mobo: gigabyte p35 ds3r
Gfx: 320mb evga 8800gts
RAM: 2gb crucial ballistix at 666 mhz 4-3-4-12 (stock can't really remember the timings)
I got:
2217 - Rendering 1 CPU
5198 - Rendernig X CPU
1.91 - Multiprocesser speedup
OpenGL - 117 (The opengl one is really low...)
Edit: NVM! Just realized there's no OpenGL support for Windows 7. What about the CPU scores?
Btw I"m on windows 7 is that maybe why...
crazlunatic, you need to post scrteenshots when posting benches. Just click "Run All Tests". Run it at default settings. You can OC if you like. That's up to you how far you wanna push it. Hope that helps you. BTW, Win7 is a beta. There's no real reason to run benches on beta's as there is nothing to compare them to. They will have bugs with no fixes, and any comparisons drawn would just be guessing. I;d install XP or Vista or both and run these benches. You want to bench with release software. Beta's could cause anything.
Now here is one I just did, and wow are those 8 threads a beast with this program...
CINEBENCH R10
************************************************** **
Tester : shane
Processor : Intel(R) Xeon(R) CPU X5450 @ 3.00GHz
MHz : 3.6
Number of CPUs : 8
Operating System : WINDOWS 32 BIT 6.0.6001
Graphics Card : ATI Radeon HD 4800 Series
Resolution : <fill this out>
Color Depth : <fill this out>
************************************************** **
Rendering (Single CPU): 8990 CB-CPU
Rendering (Multiple CPU): 56769 CB-CPU
Multiprocessor Speedup: 6.31
Shading (OpenGL Standard) : --- CB-GFX
************************************************** **
Thats with the R11 Engine,
this is base r10
CINEBENCH R10
************************************************** **
Tester : Shane
Processor : Intel(R) Xeon(R) CPU X5450 @ 3.00GHz
MHz : 3.6
Number of CPUs : 8
Operating System : WINDOWS 64 BIT 6.0.6001
Graphics Card : ATI Radeon HD 4800 Series
Resolution : <fill this out>
Color Depth : <fill this out>
************************************************** **
Rendering (Single CPU): 4424 CB-CPU
Rendering (Multiple CPU): 25876 CB-CPU
Multiprocessor Speedup: 5.85
Shading (OpenGL Standard) : --- CB-GFX
************************************************** **
so to compare, r11 32bit is almost double what r10 is in 64bit.
Competing in "The Raptor Bowl", over at The Raptor Pit, and came up with this last night on water cooling...
http://i161.photobucket.com/albums/t...hallen/CB4.jpg
Here is mine:
I may as well jump in here with my Quad !!!!
4.2 470fsb and 2gb of ram at 470 4-4-4-12
I have to retest now i have some GOOD RAM
http://img.photobucket.com/albums/v1...lmcwade/43.jpg
edit of topic but i see my 2900xt is scoring higher in the openGL everone else's cards and there is 280gtx's in some of the rigs whats gives there !!!!!
Here's mine. :D
this is whit 2 qx9775 @ stock
^^^ Um, this section is call "Xtreme Benchmark" - please keep stock scores out of here (regardless how impressive they look) ;)
Here's my Phenom II 940 @ 3.75Ghz (NB @ 2500, ram at 500Mhz 4-5-5-12).
--Matt
HMM, looks like Rol-Co's 32,282 is the number to beat yes?:D
Impressive Rol-Col, you beat me by a few points ;)
Although it shows how inefficient ST is compared to high FSB constellations.. X5460's and I would beat that easily.
Ok, here goes:
Dual X5450 (Harpertown):
http://database.he-computer.de/Bilde...finalX5450.jpg
Dual X5570 (Gainestown):
http://database.he-computer.de/Bilde...X5570_CB10.jpg
Single i7 975:
http://database.he-computer.de/Bilde...10_backup2.jpg
*ducks away before Francois comes around* :rolleyes: :p:
No, i didn't mean you as cheating, i mean there is a easy way to cheat in cinebench. But i'm not going to get started on that. You can spot the awkwards score pretty easy with the cpu's they run though.
Nah, i don't have it beat, i was just pumping him up for a better run..... Can't overclock my i7 hardly around 4ghrz, zalman 9500 starts to smell like beacon cooking.
Oh and plus a stick of my kingston hyper x t1 died, so no triple channel anymore either, go me !!!
CB11 renderer cheats are easy to spot. No secret there either, no use in creating one :rolleyes:
See if you can beat any of my scores eh? ;)
Thanks Dave.
But what you got coming up will set the record for now.. I predict at least 35K ;)
Edit: Oh yeah, it's close to 28k actually :D
The MT speedup is interesting. Back on my 9850, I used to note 3.95x or so. On the 940 I only get like 3.6x.
Some folks need to update their cpu-z
http://www.cpuid.com/download/cpuz/cpuz_151_setup.exe
or
http://www.filehippo.com/download_cpuz/
beta 1.52
My Q9650 @ 3.9Ghz: (15320)
http://i619.photobucket.com/albums/t...inebench10.jpg
I've gotta get a multi-socket setup going...
These are my scores since I got my Phenom II in January. Most are on the 0905 BIOS on this M3A79-T Deluxe motherboard, but two are with the new 1102 BIOS and Catalyst 9.7 drivers. My CPU is clocked at 3.2Ghz Core/2.4Ghz Northbridge for all tests. Also, I have a brief description of the Hyper Transport options I had enabled or disabled. From those tests I can conclude that with this motherboard, the 905 BIOS, Catalyst 9.7 drivers, and ram with a high clock and low timings will have the highest performance for this particular setup. My OpenGL scores blow all of you away with this stock clocked HD4970 :yepp:.
http://i145.photobucket.com/albums/r...BR10Scores.jpg
http://i161.photobucket.com/albums/t...allen/CB01.jpg
CPU was actually at 5195.6MHz, but CPU-Z bugged and dropped the multi from x31 to x30
Awesome.. LN2?
Still can't beat my Dualie tho :D
You know what? i'll beat myself :D here you are .
http://rol-co.nl/xs/cinebench10st.jpg
got it a long time ago already , thats why the cpu-z is outdated :ROTF:
Especialy for Dave :p:
http://images.hardwareinfo.net/users...s/009844-7.jpg
Dammit.. you beat me to it, I wanted to have 40k first :/
What cooling? If air/water I am amazed that your CPU survived...
Ah, thought so.. still, VERY nice results Rol-CO!
Maybe I should test this Q3QR on SS as well...? :D
But my RIIE is a POS, tops out at 212-213 BCLK (yes, it's definitely a board limit/bug).
Gotta wait for the Classified to support them...
2.4GHz Gulftown @ 4.57GHz, limited by max BCLK on my RIIE :(
http://www.abload.de/img/240bclk-cinebencho5bk.png
Got it at 235 here chri$h , you get a nice memspeed with that and bring back the i7 performance ,
notish that running gulftown the nb getting much hotter, you have the same issue ?
Frankly, i hate the unreliability of Cinebench R10 to properly show the performance of the CPU.
I ran the test 10-12 times in a row and look what results i got. Totally wack. The error margin is way to big in my opinion.
I can find an almost 800 points difference between 2 results ran at exactly the same settings and same load, i didn't touch the PC while running each of the tests.
Forgot to put a CPU-Z in there, i was running the 920 at 3.486ghz, DDR3 running at 1660, CL9 (OC from 1333)
Thats not cinebench but unstable system or trottling cpu .
No, it's not an unstable system. I ran 4 hours of prime 95, a 12 hour 3dsmax mental ray rendering and lots of benches, never did it even hick-up on me.
My temps never go above 67-68 in full load, even in prime 95, so no throttling because of bad cooling (mugen2 with 2 enermax magma in push/pull, good case airflow as well, CM690 as case, good PSU as well).
Usually, the load temps in normal applications are like 58-60.
It's the software, not the system.
Plus, i personally made a 3dsmax test scene, which i use to bench cpus, renders with Mental Ray.
Ran it 4 times at the exact same setup and clocks, the results were 10min29sec, 10min32sec, 10min33sec, 10min33sec.
So pretty much it gave me really stable results, not having huge error margins. While cinebench, well, you can look in the screenshot for what it does.
He's right guys.. CB10 gives different results every time you run it. That's why you gotta run 10-20x to get a good score.
http://i207.photobucket.com/albums/b...ebench4180.jpg
220x19 on i7 920 HT on
~Bex
This is mine
Q9650@ 4680 / GSKILL PI 9600 2x2 @ 1249 PL8
http://img704.imageshack.us/img704/6...ebench4680.png
Cheers and Happy 2010 !!! :up:
What batch is that? Usually they can do 230+.. must be your mobo. My RIIE couldn't pass 211/212 either no matter what, lol. Some are just bad I guess.
stock clocks, nothing fancy..
http://farm5.static.flickr.com/4007/...79ea5c3a_o.png
24/7 clocks.
--Matt
My Cinebech 10 64bit - 16 Thread run (dual Gainestown cores, stock clocks):
http://img203.imageshack.us/img203/1167/cinebench10.jpg