CIne10 and PovRay are toy benchmarks and have zero importance(name me one accepted and widely used app based on povray and cine10 engine).
The rest is OK and a healthy bump in both perf. and lower power draw.
thanks for posting this :)
Printable View
CIne10 and PovRay are toy benchmarks and have zero importance(name me one accepted and widely used app based on povray and cine10 engine).
The rest is OK and a healthy bump in both perf. and lower power draw.
thanks for posting this :)
It's 7.3% faster clock for clock versus the 2.3GHz Opteron 2356 and Cinebench scales almost perfectly with clockspeed. Is your Agena score with 64-bit Cinebench?Quote:
Originally Posted by Titan7171
AMD Opteron 2384 2.7ghz 45nm
AMD Opteron 2356 2.3ghz 65nm
Is this correct? Also what speed does the Xeon L540 run at?
2.66ghz 45nmQuote:
Originally Posted by gallag
you can still compare your own systems performance. I dont have alot of those benchmarks so this is the one Im using for my own reference, thats why Im comparing my score.Quote:
Originally Posted by informal
shanghai and barcelona are server. agena and deneb are desktop.Quote:
Originally Posted by Titan7171
That's with an app that showed nearly no improvement going from K8 to K10(single core scores).It uses scalar SSE to benchmark the CPU and is intel "tuned"(they even use it to show off 45nm Penryn and Nehalem gen if chips)Quote:
Originally Posted by accord99
I know this but are the clocks/test results comparable between desktop and server?Quote:
Originally Posted by roofsniper
im not sure but i wouldn't say so because of different ram and because the opteron chips are better than phenoms. but really i have no idea.Quote:
Originally Posted by Titan7171
ok cool I guess someone who knows will chime in.......off to play some left4dead:p:Quote:
Originally Posted by roofsniper
Opteron systems use slower memory. They top out with DDR2-667 ECC Registered.
The Shanghai core Opterons will support the DDR2-800 ECC Registered standard.Quote:
Originally Posted by Particle
http://www.newegg.com/Product/Produc...C%20registered
ECC registered is slow compared to desktop. at 800mhz the timings are like 5-5-5-18 and 6-6-6-24.Quote:
Originally Posted by Oliverda
error correcting check < thats what ECC (I think)
yea ecc registered ram is built for stability.Quote:
Originally Posted by demonkevy666
I am using Cinema 4D as of now. :rofl: But I'm using the vRay renderer so it doesn't count :)Quote:
Originally Posted by informal
opteron will be clocked slower as well, or deneb will be clocked higher then shanghai the other way round. remember the thermal stress in server racks and the much higher need for reliability in the server section..
:ROTF:Quote:
Originally Posted by informal
That's the best one I've seen yet. :up:
Quote:
Originally Posted by Particle
Quote:
Originally Posted by Oliverda
Quote:
Originally Posted by demonkevy666
Quote:
Originally Posted by roofsniper
so because of all these reasons you could say that Opterons are kinda krippled just to be server stable.Quote:
Originally Posted by Oese
well between the memory increase going from 667/800 to 1066 for Deneb, the clock for clock increase, the increase in total clocks (2.7 vs 3.0ghz) + overclocking we should see very nice increases overall:yepp:
there where benchmarks done with Barcelona and 8800 GTX on a sever with 677mhz ECC.Quote:
Originally Posted by Titan7171
http://www.anandtech.com/printarticle.aspx?i=3092
it us to have a 3dmark numbers too. :/
I think it scored around 9,00 points and a phenom scored around 9,800 10,250.
The TDP is only 75w;) I wonder what stock vcore is for these CPUs:p:
Wonder what kind of speeds they would get out of a 140w Deneb:)
Just to let people know that memory isn't that crucial with bigger CPU caches. And with wider memory controller the timings become more important rather than the frequency. Also, 800 MHz CAS 4 is slower(only due to the lower bandwidth available, actual CAS latency stays the same) than 1200MHz with CAS 6 or 1600 MHz with CAS 8.Quote:
Originally Posted by demonkevy666
Yet this is their first launched 45nm batch, thus I assume they have room to improve, quite a bit.Quote:
Originally Posted by knopflerbruce
Of course, a dual-socket system also has the benefit of running two memory controllers. One bank for each CPU generally.
informal, I agree povray is a bit of a toy benchmark, but CineBench?
I thought Cinebench was from one of the most popular rendering programmes?
If so I wouldn't call it minor. Optimising for SSE isnt just optimising for intel either, the Opterons do run certains subsets of of SSE faster than their intel equivalents(or they definitely did vs P4 even though intel came up with them, not sure about Conroe etc)AFAIK.
Quote:
Originally Posted by knopflerbruce
I believe that the 75W number is actually for ACP so true TDP is something like 110W or something like that.
105W ACP = 125W TDPQuote:
Originally Posted by qurious63ss
75W ACP = 95W TDP
55W ACP = 68W TDP