![[ch]](/mwf/v12/flags/ch.png "Switzerland")
Hi all,
I know this is not totally right but I bought a few Years back an i7 980x and it seems that when OC its still to notch exept the servers which are not in the same category which can outperform this chip.
But besides that, it doesn't that hardware computer did a big progress since a few Years since the latest chips i7-3960 doesn't seem to bring a real progress.
I used to thought that every 18 months, speed would be x2 ?
rgds
I know this is not totally right but I bought a few Years back an i7 980x and it seems that when OC its still to notch exept the servers which are not in the same category which can outperform this chip.
But besides that, it doesn't that hardware computer did a big progress since a few Years since the latest chips i7-3960 doesn't seem to bring a real progress.
I used to thought that every 18 months, speed would be x2 ?
rgds
![[in]](/mwf/v12/flags/in.png "India")
Running an I7 3930 K @ 4.8 GHz here. Very happy with its performance... was running an I7 920 earlier.
Coming to your question, I feel that the problem with progress is not so much with the Hardware as with the Software.
For instance, I'm consistently held to draws while using Houdini 2.0c and playing with a friend running an I7 2600K.
At the same time, I'm also able to get several draws with another guy running dual Xeons !!
To put it simply, I feel that even the best Software, at this point in time, is simply not able to make the best possible use of the cutting-edge Hardware offering higher kn/s speeds.
Hardware speed DOES make a difference, but not as much as one could hope for !
Hopefully, this will change once Houdini 3 and Rybka 5 are released !
Coming to your question, I feel that the problem with progress is not so much with the Hardware as with the Software.
For instance, I'm consistently held to draws while using Houdini 2.0c and playing with a friend running an I7 2600K.
At the same time, I'm also able to get several draws with another guy running dual Xeons !!
To put it simply, I feel that even the best Software, at this point in time, is simply not able to make the best possible use of the cutting-edge Hardware offering higher kn/s speeds.
Hardware speed DOES make a difference, but not as much as one could hope for !
Hopefully, this will change once Houdini 3 and Rybka 5 are released !
White it is likely that when ELO of engines is improving, the more draws you may see, if you simply compare the speed of these new CPUs, they haven't really made progress compared the the old released i7-980x
Is it related to the fact that Intel does not have anymore competition ? I dont know....
Rgds
Is it related to the fact that Intel does not have anymore competition ? I dont know....
Rgds
Shrapnel:
Looking at comparable (or better, the same middlegame positions) how much faster in Kn/s do you get with i7 3930 @ 4.8 Ghz. vs i7 920 (at what speed?).
(I have i7 930 at 3.88 Kn/s.)
Thanks, CMA
Looking at comparable (or better, the same middlegame positions) how much faster in Kn/s do you get with i7 3930 @ 4.8 Ghz. vs i7 920 (at what speed?).
(I have i7 930 at 3.88 Kn/s.)
Thanks, CMA
![[hr]](/mwf/v12/flags/hr.png "Croatia")
It will soon be 4 years that I've got my q6600, and I still don't find it feasible to upgrade it to something new (feasible = invest a monthly wage for at least 3x speedup). I thought it was going to happen in a year or two, now it is double that time and I'm still doing 275 kN/s with R4.1 not having true reason to upgrade (not playing on playchess either). So, I agree with you Bouddha, the progress has been somewhat slow lately.
Edit: of course, I'm talking budget versions.
Edit: of course, I'm talking budget versions.
![[ru]](/mwf/v12/flags/ru.png "Russian Federation")
So as I am. Upgraded to Q6600 and don't want anything less than 16-core 128-bit CPU.
![[ca]](/mwf/v12/flags/ca.png "Canada")
I have been looking at options but with revenue way down all I can do right now is look
![[de]](/mwf/v12/flags/de.png "Germany")
If we look at Intel's Tick Tock Model http://www.intel.com/content/www/us/en/silicon-innovations/intel-tick-tock-model-general.html, the leaps forward for the Tocks seem to get much smaller.
Pentium 4 to Core 2: a huge jump in integer performance
Core 2 to Nehalem: ~+30% jump in integer performance and a huge jump in memory performance
Nehalem to Sandy Bridge: nothing really new (OK, a little bit more memory performance and lower idle power consumption)
This is the impact of the Ticks:
Xeon 5300 (65 nm) to 5400 (45 nm): 13% higher speed
Nehalem (45 nm) to Westmere (32 nm): 50% more cores
But first reports of Ivy Bridge are not really promising.
So to me it really looks like there is some slowdown.
Pentium 4 to Core 2: a huge jump in integer performance
Core 2 to Nehalem: ~+30% jump in integer performance and a huge jump in memory performance
Nehalem to Sandy Bridge: nothing really new (OK, a little bit more memory performance and lower idle power consumption)
This is the impact of the Ticks:
Xeon 5300 (65 nm) to 5400 (45 nm): 13% higher speed
Nehalem (45 nm) to Westmere (32 nm): 50% more cores
But first reports of Ivy Bridge are not really promising.
So to me it really looks like there is some slowdown.
So that mean you invested just in time with your cluster and should be safe for a while.
Lukas:
Wasn't there a report of the Ivy Bridge at 7.0 Ghz/s ?
As I remember, that was on dry ice, but even so, oshouldn't the ordinary O/C system n water be able to get > 5.0 Ghz/s?
Wasn't there a report of the Ivy Bridge at 7.0 Ghz/s ?
As I remember, that was on dry ice, but even so, oshouldn't the ordinary O/C system n water be able to get > 5.0 Ghz/s?
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill