1
2
![[be]](/mwf/v12/flags/be.png "Belgium")
Just a great configuration that everybody can dream from!
Things i would do different..for sure i would take a other power supply..where i have the exhaust from the psu to the back..now the warm air blows into the case!
And he can simple at two fan's more to get a push pull configuration..win a few degrees!
JP.
Things i would do different..for sure i would take a other power supply..where i have the exhaust from the psu to the back..now the warm air blows into the case!
And he can simple at two fan's more to get a push pull configuration..win a few degrees!
JP.
![[mx]](/mwf/v12/flags/mx.png "Mexico")
>PC Pr0n photos
Hi,
Now at 4.2Ghz ... Sorry pics are not to great had to take em with a camera
Zap Mex at over 10500 KNS
Rybka 4 without large pages
Regards
Now at 4.2Ghz ... Sorry pics are not to great had to take em with a camera
Zap Mex at over 10500 KNS
Rybka 4 without large pages
Regards
![[us]](/mwf/v12/flags/us.png "United States")
What did you use for thermal paste? No mouse pad ? : )
Wow, 917. What do you get in Rybka 3?
Would be interesting to compare it with Sedat's benchmark. Seems like the 40-core cluster is still a ways away but these cpus are catching up all the time.
Yes I am aware of Lukas' upgrades since then..
Wow, 917. What do you get in Rybka 3?
Would be interesting to compare it with Sedat's benchmark. Seems like the 40-core cluster is still a ways away but these cpus are catching up all the time.
Yes I am aware of Lukas' upgrades since then..
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
my present Skully 8 cpu at 3.6Ghz gives 397 kN/s in same poition with R4
and with Zappa 4.800 kN/s
and with Zappa 4.800 kN/s
*GASP*
It seems the new computer will be 3 times as fast as a skully, that's the equivalent of 24 cores! So each core is like twice the speed
It seems the new computer will be 3 times as fast as a skully, that's the equivalent of 24 cores! So each core is like twice the speed
although i am no expert i believe Skully was throttled by having to use FB Dimms in memory.
I get 536 with my 980 at 4.02
>I get 536 with my 980 at 4.02
Hi,
I get 710 at 4.5Ghz with my 980
Regards
Ahh, so for the 980 at 4020 mhz you can get 536 kn/s, whereas at 4500 mhz you can get 710 kn/s.
The first one is ~0.13333 kn/s per mhz, and next one is ~0.16 kn/s per mhz. An increasing return on clockspeed? Well I suppose the cpu clock isn't the only factor.
That's ~0.025 kn/s per mhz every increase of 480 mhz. At this rate 9300 mhz (over 9000) will yield a whopping 3813 kn/s!
Surely there are no serious hurdles in overclocking a 980 that far? :o)
The first one is ~0.13333 kn/s per mhz, and next one is ~0.16 kn/s per mhz. An increasing return on clockspeed? Well I suppose the cpu clock isn't the only factor.
That's ~0.025 kn/s per mhz every increase of 480 mhz. At this rate 9300 mhz (over 9000) will yield a whopping 3813 kn/s!
Surely there are no serious hurdles in overclocking a 980 that far? :o)
Hi,
Not sure what you are saying? the only other thing I can add is that I had LP running although I must admit that tends to be hit or miss for me whether the KNS goes up or down
http://www.hiarcs.net/forums/viewtopic.php?t=3619
Regards
Not sure what you are saying? the only other thing I can add is that I had LP running although I must admit that tends to be hit or miss for me whether the KNS goes up or down
http://www.hiarcs.net/forums/viewtopic.php?t=3619
Regards
> Not sure what you are saying?
He's saying that the nodes per second don't increase linearly with hertz added, they increase higher than that, so that if you doubled the computer speed, the nodes would multiply by 5.
It could be an error on Vas's node recalculation, but there is evidence that the stronger the computer is, the higher Rybka overperforms (i.e. also not linear, I think it's a bug that weakens Rybka at slow hardware, it has yet to be tested).
Hmmm, I need to try 4.5 looks like.
>Hmmm, I need to try 4.5 looks like
When I get back from Cuba I am gonna try 5Ghz
Regards
What mobo are you useing?
>What mobo are you useing
For the 980X a Asus Rampage Extreme 3
Regards
the best in my opinion yet, not to many use it in the engine room..why is that?
>the best in my opinion yet, not to many use it in the engine room..why is that?
Hi,
That is a tough one to answer, it may simply be that others (and I assume we are talking about the Asus Rampage Extreme 3) bought there motherboards before the ARE3 hit the market. I have not played in the engine room for some 2 years so am not in a position to know what setups people are running
Regards
Tony
Because you can kibitz just as well with your macro running on a P4 machine!
>Because you can kibitz just as well with your macro running on a P4 machine!
I wish I could kibitz (I think) the decent games with a macro but sadly I have to collect chess games like everyone else, unless you have a macro and instructions on how to use it you want to send me On another note as I am a Chessbase 10 user the makers certainly don`t help owners with large databases ... The new CB11 seems to have a some new fluffy bits but no news on speeding up of database searches, sorting, removal of doubles etc etc etc
Regards
I get 536 with my 980 at 4.02
even with my lesser hardware i can still play good chess.
even with my lesser hardware i can still play good chess.
![[kw]](/mwf/v12/flags/kw.png "Kuwait")
I was wondering what had happened to that build. Personally I would have gone for water cooling, and for the case I would have chosen the Thermaltake monster case which has much more real estate and would be much easier to fit stuff in (especially if you move to water). Anyway the switch to water could come at a later stage and is not too difficult. Have you tried testing the motherboard on phase? If these processors + motherboard combo can get to 5 Ghz then it might become irresistible. Good luck with the build!
By the way if you are using Win 7 you can use the built in utility to clip an image of the screen (it is the pair of scissors). Otherwise you have Hypersnap which is also very good.
By the way if you are using Win 7 you can use the built in utility to clip an image of the screen (it is the pair of scissors). Otherwise you have Hypersnap which is also very good.
>Have you tried testing the motherboard on phase? If these processors + motherboard combo can get to 5 Ghz then it might become irresistible.
Hi Madj,
It is not my comp I am only building it for Paul (NATIONAL12) but personally I believe 5Ghz is more than poss on phase maybe even on a waterchiller.
You have to remember this is Pauls comp and I will not go crazy with the voltages especially on air. I use snagit 10 on my comp to do screen captures but i am still on XP64, I really ought to get up to date with windows 7
>Well if you can handle and not think to much as a chess player..run this test for me and see if you can compare with the big guys
Hi jpqy,
with the talk of big guys I thought you meant down the gym deadlifting 300Kg lol
....Give me a little time to finish tweaking the memory the latencies are still poor the BIOS does not have the options of my Asus Rampage 3 Extreme http://www.hiarcs.net/forums/viewtopic.php?t=3619 Regards
Tony
Hmmm ... I might have to give this a try. I will probably wait for the 8 core CPU to come out though as that seems very close now. How is stability on this motherboard? Have you tried running a tourney overnight with it overclocked. I have had my Skulltrail overclocked at 4 Ghz for oh so long now, and it is super solid and the most reliable and stable system I have ever had. I had it overclocked at 4.8 Ghz on phase for quite some time, but to be honest that was way too noisy and not practical. I would have to leave the unit running in the basement in a locked room. Since moving to water, I can use this computer as an everyday computer and it is a joy to use for Photoshop or if I want to edit some videos. I think it is about time I upgraded it and if this is very stable I might go for it.
The next computer I'm saving up for is twin 8 core Bulldozers. I think there will be an 8 core version in the sever range.
*dreaming*
*dreaming*
They will probably have 16 core Bulldozers :)
2 or 4 of those should be sweet.
http://www.infoworld.com/d/hardware/amd-details-next-gen-architecture-16-core-server-chips-017
2 or 4 of those should be sweet.
http://www.infoworld.com/d/hardware/amd-details-next-gen-architecture-16-core-server-chips-017
Hi,
Your reasons for moving away from phase cooling are basically the same as mine .... I updated the BIOS to the A47 one which added a few new options and is supposed to help memory OC ...
An option in the BIOS is EXTREME COOLING which has 2 modes, the first for below 0C and the second for below -50C but that is all it says and does not give any clues t what it does, maybe it increases voltages across the board? it was like having a red button in front of you with `do not press` printed on it and you know you want to press it lmao but I resisted ... I did test that Arena2+cyclone xtreme bench thingy but the result was the same at stock speed as it was at 4.2Ghz I must have done something wrong I may return to that problem later all other benchies increase dramatically.
The new BIOS has the option to play with UNCORE which I hope will get rid off the poor memory latencies
Regards
Your reasons for moving away from phase cooling are basically the same as mine
.... I updated the BIOS to the A47 one which added a few new options and is supposed to help memory OC ...An option in the BIOS is EXTREME COOLING which has 2 modes, the first for below 0C and the second for below -50C but that is all it says and does not give any clues t what it does, maybe it increases voltages across the board? it was like having a red button in front of you with `do not press` printed on it and you know you want to press it lmao
but I resisted ... I did test that Arena2+cyclone xtreme bench thingy but the result was the same at stock speed as it was at 4.2Ghz I must have done something wrong I may return to that problem later all other benchies increase dramatically.The new BIOS has the option to play with UNCORE which I hope will get rid off the poor memory latencies
Regards
I actually screwed up a couple of Skulltrails, but I don't think it was because I was pushing too hard. On one of them I simply put some grease designed for sub zero environments in the socket, which started to boil and pop when I tried to use it at default speeds with water cooling. I had run out of simple silicone heat compound and figured that would be better than Vaseline. The other was probably a bad motherboard as it simply would not read all the memory modules when fully loaded. Best advice is to forget about the advice Vapochill gives regarding filling the CPU socket with anything, as condensation turns out to not be as big a problem inside the socket as most think. I still have a system running perfectly at 4 Ghz after it ran for a good 6 months on Vapor at 4.8 Ghz. Giving up .8 Ghz for the usability of a water cooled system is definitely worth it. I guess it would take me a couple of hours work to set it up on vapor again.
The main problem with the Skulltrail was that it was using FBDIMMS which get incredibly hot. I mean you can singe your fingers if you touch the memory and it is not actively cooled. I did have a plan to super cool the memory and MCH, but got interested in other things. My guess is I could have pulled in another .4 Ghz out of the system and gotten 5.2 Ghz stable. I could get it to 5.2 Ghz but it would crash Prime 95 although it could run the Fritzmark test successfully. With non FBDIMMS I think a large limiting factor is gone. You have to remember that when you vapor cool a CPU, you are actually reducing dramatically the pathways of electrons. That means much less "friction" when electrons move around as they have much more theoretical space. So at default voltages you should find that you can run the system at dramatically faster speeds simply because there is less resistance for the electrons. Of course you can push things to a higher level by pushing voltages, but then you have to keep an eye on heat. Maybe if you watercool the MCH (if it has something like that) and the memory you can get better performance, but then things start to get quite complicated. I think watercooling in general is quite robust if you have the right parts, but you really have to have experience in setting it up properly and have proper ingredients in the coolant and the proper piping and HS (avoid acrylic at all costs) to avoid eventual fouling.
The main problem with the Skulltrail was that it was using FBDIMMS which get incredibly hot. I mean you can singe your fingers if you touch the memory and it is not actively cooled. I did have a plan to super cool the memory and MCH, but got interested in other things. My guess is I could have pulled in another .4 Ghz out of the system and gotten 5.2 Ghz stable. I could get it to 5.2 Ghz but it would crash Prime 95 although it could run the Fritzmark test successfully. With non FBDIMMS I think a large limiting factor is gone. You have to remember that when you vapor cool a CPU, you are actually reducing dramatically the pathways of electrons. That means much less "friction" when electrons move around as they have much more theoretical space. So at default voltages you should find that you can run the system at dramatically faster speeds simply because there is less resistance for the electrons. Of course you can push things to a higher level by pushing voltages, but then you have to keep an eye on heat. Maybe if you watercool the MCH (if it has something like that) and the memory you can get better performance, but then things start to get quite complicated. I think watercooling in general is quite robust if you have the right parts, but you really have to have experience in setting it up properly and have proper ingredients in the coolant and the proper piping and HS (avoid acrylic at all costs) to avoid eventual fouling.
cooling the FB DIMMS has always been a problem to me on air.i cant remember what i am using but they are huge things with multiple fins,likewise i am unhappy with Zalman 9700 coolers.i am sure a few updates could improve matters.however i should not complain it's ran for 2 years now often 24/7 for weeks with only one failure with memory which was replaced free under guarantee.overall very reliable(ran o/c at 3.6).
have u considered the self contained koolance.systems..
just thought i would mention this.the monitor is 24" so it gives you some idea how big this is.
Thanks for the clarification because the chair, mouse and keyboard could all be the size of a mountain.
Smart ass!
If you hit the prt screen button.. it copies the screenshot to the clipboard.
![[ee]](/mwf/v12/flags/ee.png "Estonia")
How much did all this cost altogether?
about £4500 inc 28" monitor.
![[nl]](/mwf/v12/flags/nl.png "Netherlands")
What a beast, congrats
What is your memory setup? 6x2 GB?.
What is your memory setup? 6x2 GB?.
yes.1866 MHz.
>What did you use for thermal paste? No mouse pad ? : )
Hi,
Thermal paste is Artic Silver 5 ... No need for a mouse pad as it works fine and I am only testing.
Got to tweak the memory as the latencies are poor ATM and although I have not put any pics up the Everest CPU benchmarks are staggering
The Fritz benchmark prg is totally out of date and only benchies upto 8 cores
>What is your memory setup? 6x2 GB?.
It is not mine but NATIONAL12`s (aka Paul) machine and yes at the moment Paul has 6x2GB with 6 more mem slots free
Regards
Well if you can handle and not think to much as a chess player..run this test for me and see if you can compare with the big guys
http://www.xtremesystems.org/forums/showthread.php?t=202139
i would like to have a result with HT and without HT !!
Then you have no problem anymore with this Fritz benchmark.
Regards,
Jean-Paul.
http://www.xtremesystems.org/forums/showthread.php?t=202139
i would like to have a result with HT and without HT !!
Then you have no problem anymore with this Fritz benchmark.
Regards,
Jean-Paul.
2.5x less price-effective as mine as far as CPU speed is concerned. You must be either very rich or yo ujust value so much computer chess
i am not very rich.i look at another way less than a third of the price of buying an average new car,at least in UK and i know which i get more fun from.
Congrats and gl on your new build Paul. Great build@tony
![[br]](/mwf/v12/flags/br.png "Brazil")
I'm not too sure, but I think that when you purchased your Skulltrail, you said you
was getting old and would be the last hardware you would buy in your life.
Congratulations, long life to you, Paul!
was getting old and would be the last hardware you would buy in your life.
Congratulations, long life to you, Paul!
enjoying life to much.
so the next logical step is to go heads-up against lukas' cluster and embarrass it by holding to a long string of consecutive draws! :-)
Lukas knows i will only play him at corr time controls when i would fancy my chances against cluster.easier than playing turbojuice.
1
2
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill