- Joined
- Apr 26, 2010
- Messages
- 31,256
- Reaction score
- 13,886
- Points
- 26,065
- Age
- 34
- Location
- Norkem Park, Gauteng
I'm going to get back into LN2 in the near future (yes, I'll stop using Vaseline...) and have some hardware that I think shows a bit of potential.
1. Socket AM2
Frequency
I have an Athlon X2 6000+ that managed 4,144.57 MHz - at one stage it was the record for AM2 but things have changed and it is now 3rd for the 6000+ and 16th for AM2. I did my run using DICE and the chip hasn't been tested any colder. If it doesn't cold bug too soon, my first goal is 4,235 MHz to take first place for the 6000+, and then 4,299 MHz to take the AM2 record back.
PiFast
Using the same 6000+ I'm currently at 32.130s for PiFast, which is 6th for the 6000+ and 13th for AM2. The kicker is that I was at only 3,901 MHz with the RAM at a very low 487 MHz 4-4-4-17 2T. The record is 30.190s with a 6000+ at 3,962 MHz (only 61 MHz faster) and the RAM at 566 MHz 4-4-4-11 2T. I find it likely that my RAM can bench at around 560 MHz 3-3-3-9 2T.
SuperPi 1M
Here I've got a really sucky score of 21.780s with that same 6000+ at 3,965 MHz, which is 6th for the 6000+ and 31st for AM2. The RAM was at a low 496 MHz 4-4-4-12 2T, just tightening the timings to 3-3-3-9 should get me a much faster run. The 6000+ record is 20.781s and the AM2 record is 20.220s. I'm not very confident about this one as the record is at 4,092 MHz, but at the same time I know full well I didn't even try to tweak that run. I also know for absolute sure that the CPU can bench SuperPi 1M at over 4 GHz - that run is slower than the 3,965 MHz run as I did it with a single stick of memory which seeing what the CPU can do.
SuperPi 32M
I never really tried here. I know I can run the bench at 3,937 MHz (which resulted in a very poor time of 21m56.157/URL] - using single channel memory at a slow 487 MHz 4-4-4-17 so there's close to a minute that can be shaven off right there) - the only reason for that run was to see where it can 32M on phase change. With LN2 or DICE I can go higher, the problem here was the duration of the benchmark meant the CPU was going into the positive temperatures. Better cooling will probably put my frequency at a minimum of 3,080 MHz, which is around the speed I'll need to hit top three for AM2 (there are two monstrous 6400+ chips that did 32M at 4,080 MHz, I have severe doubts about catching them).
wPrime 32M
I can't believe I haven't even tried this yet. Someone else used my CPU to get [URL="http://hwbot.org/submission/615551_team_ghetto_wprime___32m_athlon_64_x2_6000_(windsor)_21sec_670ms"]21.670s, which is 11th place for the 6000+ and 65th for AM2. It was done at 3,901 MHz, 19.484s has been done at 3,928 MHz, and I KNOW I can bench it higher than 3,901 MHz as the person that used by CPU never even saw 4 GHz while I managed to bench higher than that. It's a maybe in my eyes.
wPrime 1024M
This chip has never done a wP1024, so /shrug
2. Socket 939
Frequency
This is a tall order I have an FX-55 that validated 3,619.49 MHz, which was 3rd for the FX-55 at the time although it is now 4th for the San Diego FX-55 core, 5th for overall FX-55, and 31st for Socket 939. First, I used under 2v and only -45'c while I have the option of up to 2.108v and I know this CPU doesn't have a cold bug, so -196'c. I had validations of over 3,960 MHz but they were corrupt Anyway, 4,000 MHz will be the first milestone (only 16 people have ever hit it), then 4,076 MHz for the San Diego FX-55 record, then 4,123 MHz for the FX-55 record, and finally 4,209 MHz for the overall Socket 939 record. I have very little confidence for this xD
PiFast
I have only run at around 3.1 GHz. I have RAM that can bench 300 MHz 2-2-2-5 which is a big advantage.
SuperPi 1M
Don't judge me, I have nothing saved for SuperPi on this CPU.
SuperPi 32M
Don't judge me, I have nothing saved for SuperPi on this CPU.
wPrime 32M
Currently I'm at 50.468s at 3,385 MHz using phase change cooling and single channel RAM (which is fine) which puts me at 7th position for the FX-55 and and don't even ask about Socket 939 as I can't compete with dual core CPUs. The FX-55 record is 40.062s at 3,815 MHz. I have a bit of faith.
wPrime 1024M
Right now I'm at 27m56.140s at 3,306 MHz using phase change cooling and single channel RAM (fine) which puts me 6th for the FX-55. Again, I can't compete with dual core CPUs so I'm looking at the single core Socket 939 record, which is 22m29.593s at 3,609 MHz. I have a bit of faith.
3. LGA775
I have no plans other than to hit 7 GHz (or even better, 8 GHz).
1. Socket AM2
Frequency
I have an Athlon X2 6000+ that managed 4,144.57 MHz - at one stage it was the record for AM2 but things have changed and it is now 3rd for the 6000+ and 16th for AM2. I did my run using DICE and the chip hasn't been tested any colder. If it doesn't cold bug too soon, my first goal is 4,235 MHz to take first place for the 6000+, and then 4,299 MHz to take the AM2 record back.
PiFast
Using the same 6000+ I'm currently at 32.130s for PiFast, which is 6th for the 6000+ and 13th for AM2. The kicker is that I was at only 3,901 MHz with the RAM at a very low 487 MHz 4-4-4-17 2T. The record is 30.190s with a 6000+ at 3,962 MHz (only 61 MHz faster) and the RAM at 566 MHz 4-4-4-11 2T. I find it likely that my RAM can bench at around 560 MHz 3-3-3-9 2T.
SuperPi 1M
Here I've got a really sucky score of 21.780s with that same 6000+ at 3,965 MHz, which is 6th for the 6000+ and 31st for AM2. The RAM was at a low 496 MHz 4-4-4-12 2T, just tightening the timings to 3-3-3-9 should get me a much faster run. The 6000+ record is 20.781s and the AM2 record is 20.220s. I'm not very confident about this one as the record is at 4,092 MHz, but at the same time I know full well I didn't even try to tweak that run. I also know for absolute sure that the CPU can bench SuperPi 1M at over 4 GHz - that run is slower than the 3,965 MHz run as I did it with a single stick of memory which seeing what the CPU can do.
SuperPi 32M
I never really tried here. I know I can run the bench at 3,937 MHz (which resulted in a very poor time of 21m56.157/URL] - using single channel memory at a slow 487 MHz 4-4-4-17 so there's close to a minute that can be shaven off right there) - the only reason for that run was to see where it can 32M on phase change. With LN2 or DICE I can go higher, the problem here was the duration of the benchmark meant the CPU was going into the positive temperatures. Better cooling will probably put my frequency at a minimum of 3,080 MHz, which is around the speed I'll need to hit top three for AM2 (there are two monstrous 6400+ chips that did 32M at 4,080 MHz, I have severe doubts about catching them).
wPrime 32M
I can't believe I haven't even tried this yet. Someone else used my CPU to get [URL="http://hwbot.org/submission/615551_team_ghetto_wprime___32m_athlon_64_x2_6000_(windsor)_21sec_670ms"]21.670s, which is 11th place for the 6000+ and 65th for AM2. It was done at 3,901 MHz, 19.484s has been done at 3,928 MHz, and I KNOW I can bench it higher than 3,901 MHz as the person that used by CPU never even saw 4 GHz while I managed to bench higher than that. It's a maybe in my eyes.
wPrime 1024M
This chip has never done a wP1024, so /shrug
2. Socket 939
Frequency
This is a tall order I have an FX-55 that validated 3,619.49 MHz, which was 3rd for the FX-55 at the time although it is now 4th for the San Diego FX-55 core, 5th for overall FX-55, and 31st for Socket 939. First, I used under 2v and only -45'c while I have the option of up to 2.108v and I know this CPU doesn't have a cold bug, so -196'c. I had validations of over 3,960 MHz but they were corrupt Anyway, 4,000 MHz will be the first milestone (only 16 people have ever hit it), then 4,076 MHz for the San Diego FX-55 record, then 4,123 MHz for the FX-55 record, and finally 4,209 MHz for the overall Socket 939 record. I have very little confidence for this xD
PiFast
I have only run at around 3.1 GHz. I have RAM that can bench 300 MHz 2-2-2-5 which is a big advantage.
SuperPi 1M
Don't judge me, I have nothing saved for SuperPi on this CPU.
SuperPi 32M
Don't judge me, I have nothing saved for SuperPi on this CPU.
wPrime 32M
Currently I'm at 50.468s at 3,385 MHz using phase change cooling and single channel RAM (which is fine) which puts me at 7th position for the FX-55 and and don't even ask about Socket 939 as I can't compete with dual core CPUs. The FX-55 record is 40.062s at 3,815 MHz. I have a bit of faith.
wPrime 1024M
Right now I'm at 27m56.140s at 3,306 MHz using phase change cooling and single channel RAM (fine) which puts me 6th for the FX-55. Again, I can't compete with dual core CPUs so I'm looking at the single core Socket 939 record, which is 22m29.593s at 3,609 MHz. I have a bit of faith.
3. LGA775
I have no plans other than to hit 7 GHz (or even better, 8 GHz).