It uses doubledouble arithmetic by splitting the precision into two doubles. But that's only for a final division with high precision. Double precision performance is more important for calculations...
Type: Posts; User: _mat_; Keyword(s):
It uses doubledouble arithmetic by splitting the precision into two doubles. But that's only for a final division with high precision. Double precision performance is more important for calculations...
Hey poke, I've been watching your y-cruncher project for some time now. I even used it to externally validate my hexadecimal results for 32B. Your kind words are very much appreciated!
This 290X on Haswell-E is currently the fasted card in the bench: http://hwbot.org/submission/2673063_nata_58_gpupi___1b_radeon_r9_290x_19sec_690ms
I wonder if the CPU somehow helps achieving the...
First of all it's a completely different algorithm to calculate pi. It uses the BPP formula, the only pi formula that can be computed in parallel. That said it only crunches 9 digits from 1M, exactly...
Currently SLI and Crossfire are not possible, please have a look at the FAQ. But it's definitely possible in the future, I just need multiple cards to develop and implement it. Let's see if the...
Yes, the 290 seems very strong even tough it has less compute units. I don't own any R9 cards, so I can't really tell you what is going on there. But there must be good reason for this, maybe driver...
I already have a CUDA version, I implemented it before porting to OpenCL. CUDA has the nicer toolkit and much easier to develop an idea. But the goal was to let NVIDIA and AMD compete in the same...
GPUPI - SuperPI on the GPU - Post results!
https://www.overclockers.at/attachment.php?attachmentid=198154&thumb=450
In the last weeks I've implemented GPUPI, a benchmark that computes pi in parallel via OpenCL. It was something that I...
Sure, I will start a new one!
Edit: Here it is: http://www.xtremesystems.org/forums/showthread.php?289849-GPUPI-SuperPI-on-the-GPU-Post-results!&p=5242169&viewfull=1#post5242169
Some parts use double precision, but most of it is integer.
felix_w, nice to see some FirePro scores. :)
Come back to life, sweet thread ...
https://www.overclockers.at/attachment.php?attachmentid=198154&thumb=250
In the last weeks I've implemented GPUPI, a benchmark that computes pi in parallel...
We couldn't get the bclock working on a very slim Windows XP SP3. We had to install the ME driver (9.0.10 was okay for us), maybe the .net Framework 3.5 was important too (not sure), but really...
This week splmann visited me in Vienna and we benched for three days until the bitter end! :toast:
http://www.overclockers.at/attachment.php?attachmentid=187156&thumb=400...
Good work, mate!
There is another possibility. K10Stat can raise voltage to. Just set 1.55 for 1.95 volts.
Ok, I found the solution. It's really easy, maybe too easy. :)
There is a slider in Easy Tune 6, that is not limited at 1.8 volts. I will let you know, if it's working. CPU-Z is positive, but who...
Hey guys,
I'm currently benching a llano cpu with the GA-A75-UD4H board and I'm trying to get the vcore above the default 1.8 volts. Any ideas if there is a voltmod or a software trick to increase...
Nice! Especially loving the high quality photos. :)
0 °C => b3
5 °C => b5
10 °C => b7
> 15 °C boots with 2400 MHz. But IMC has to be kept warm otherwise the system crashes immediately. :(
Thanks for your help!
I tried again from scratch and now I get the b1 => 00. Temperature is at -20 to -30 °C, VTT at 1,35 V, VCSA at 1,25 V. 2133 no problem, 2400 (by bclock and strap) no chance at all. Even at CL11. Any...
I get a 00 after a 25. Is that also a sign for a weak imc? But I couldn't update to your BIOS version ("Selected file is not an efi bios"), so I am still on 070b. But I will try 075b from here.
...
Just wow! Awesome modules and sure as hell an awesome IMC. :toast:
Already tried to hurt my 3960X with some memory frequencies, but I can't get it beyond 2400 CL10. Are some IMCs really that bad?...
Are you kidding me? :D
No Lucas-Lehmer iteration was harmed here. ;)
I did not try them on ln2 for now. But they scale pretty well with voltages. 2,1 Volt was necessary for 3000 MHz CL10-13-12-31. Hope I can free some time to see if tighter timings are possible.
...
After an awful night of sleep I decided to try out a second bulldozer chip (reaches just 8 ghz on one core, no MHz more ...). It booted up with the exact same settings but scaled way better to cold...
