I can't see anything:(
If I were you, I wouldn't bet Dave on that one, you'll loose:up:
Printable View
so metriod, if intel says nehalem = 1.33 times performance of penryn (parrallellism), and Vr-zone found it to run the 3dmark bench at 1.45 times the performance on penryn, then where does your humble 23% IPC improvement estimate come from?
You're saying clock for clock Nehalem is not really a 45 to 50% improvement over Penryn, but rather closer to a 25% improvement?
Merely guestimating, would this translate into a gaming rig with a Nehalem @ ~ 3ghz performing similarly to gaming rig with a Penryn @ ~ 4ghz? (assuming the game doesnt use 4 or more threads and the rig's specs are similar/identical)?
cuz nehalem got 8 threads, two times more than yorkfield?
I'll take your money if you want to give it away..:D
I'm just a "po white boy"
I don't bet unless I know I'm going to win.
I saw that cpuz shot 2-3 weeks ago and the person that showed it to me is beyond reproach.
Let's just say one of the top people in this hobby.
...I found this all very funny here, in the end you really have to wait for the results on each single application whether it can make use of the Hyper Threading or not and the other features...
Back in time when there was a P4 with or without Hyper Threading some applications showed gain, some not. It depends on all single lines of code in the application. You can google it.
I'm coding on a Fractal Benchmark and optimized it in the recents years in assembler both for FPU and SSE2 and it's really hard to do it right for each processor architectures...as most people use C-compiler they even got less influence on it and then you have to do multi threading to make use of the Hyper Threading, what's also not so easy depending on the algoritms. In my case from the P4 results there was a gain of about 10% in SSE2 performance with Hyper Threading...still...it can be the same for Nehalem but can be also less or more as it has totally different inner architecture and execution cores...
Memory, cache and most important how SMT is handled.
Yes something like that, for me a Nehalen at 3.0Ghz = Yorkfield at 4.0Ghz, most of games use only 2 cores. So unless developers adhere to more than 2 cores, no greater benefits will be seen, a 4 cores Nehalen can make 8 threads. That is huge. I wonder how much points it can give doing Folding at Home though.
I was being funny. I sure take your word. I just imagine 16 threads executing the WCG program. Nehalem will be a breakthrough for all in the WCG community and specially for our team, just imagine how many tasks will be done with 4 cores(8 Threads) at the beginning and later on 8 cores(16 Threads), our team will reach great things, easynews will be very easy to be beaten. :rofl:
After Nehalem launches, the top will be claimed once more, reduced memory latency and SMT are all that we need.. :yepp:
Metroid.
I imagine that many members of our team (myself included) are keeping a very poised hand to click the purchase button as soon as dual socket Nehalem goodness comes on the market.
I have a very distinct feeling that around the end of this year and the first part of next we shall be seeing a nice bump in the daily output here at XSDC.
http://op-for.com/mr%20burns.jpg
I wonder if this cpu finally can handle Supreme commander...
At 1680*1050 8800GTX and QX9650 at 4.4ghz air supcom dropped to 14fps (max 107fps) in 81X81 maps that is pretty :banana::banana::banana::banana:ty and not that playable lol...