|
Post by MachineLearning on Apr 1, 2022 22:27:06 GMT -5
I didn't see a memory subsection to the OC museum, so I'm throwing this in the 2D Intel section since it was on Intel and is technically 2D. Got #1 in PYPrime 2B for DDR2 using an Intel E8500 and 2x1GB Crucial 10 year anniversary DDR2. Here's the submission: hwbot.org/submission/4964685_And, here is a link to my post on it at TechPowerUp. It has a photo of the setup, as well as all of the relevant settings I used, so if you want to see all of that then check it out: www.techpowerup.com/forums/threads/tpus-nostalgic-hardware-club.108251/post-4730683Please enjoy. Warp9 reigns supreme.
|
|
|
Post by Vinster on Apr 2, 2022 13:28:52 GMT -5
great job. we have 1st and 3rd.
Vin
|
|
|
Post by osmiumoc on Apr 3, 2022 15:26:01 GMT -5
|
|
|
Post by MachineLearning on Apr 3, 2022 19:50:05 GMT -5
Hell yeah!! I'm competitive, but I'm happy as long as the team is on top. It's far more motivating than just playing for myself. Plus I'm becoming confident in my own ability, the numbers are meaning less. On topic, I bet you can totally nail sub-30, if you can manage to get a C2D close to 5GHz... Now we got #1, #2, #4! I'll be trying again (with safe voltages ) sometime soon.....
|
|
|
Post by osmiumoc on Apr 4, 2022 20:55:26 GMT -5
Glad to hear. I had some mixed feelings about the sub, it did not feel right to snipe a team member like that. But I also wanted to make a sub to this bench. I'll probably rerun this when I make some dice runs towards the end of the current challengers DDR2 div.
|
|
|
Post by Vinster on Apr 4, 2022 23:07:36 GMT -5
Glad to hear. I had some mixed feelings about the sub, it did not feel right to snipe a team member like that. But I also wanted to make a sub to this bench. I'll probably rerun this when I make some dice runs towards the end of the current challengers DDR2 div. it's going to happen and is unavoidable to a point. Typically as we bench we look at subs for reference to get an idea what it's going to take and go for gold...
It's a touch old and I need to update it (actually didn't realize it's been 10 months crap... sorry guys.. new baby has been keeping me busy)
Vin
|
|
|
Post by Macsbeach98 on Apr 5, 2022 1:27:48 GMT -5
Dont worry Osmi I will be coming with cold towards the end of Challengers REAL COLD.
|
|
|
Post by Bones on Apr 5, 2022 6:17:45 GMT -5
Hell yeah!! I'm competitive, but I'm happy as long as the team is on top. It's far more motivating than just playing for myself. Plus I'm becoming confident in my own ability, the numbers are meaning less. On topic, I bet you can totally nail sub-30, if you can manage to get a C2D close to 5GHz... Now we got #1, #2, #4! I'll be trying again (with safe voltages ) sometime soon..... THIS. Good to see the sprit of competition driving guys to succeed. I mean it's all good in the end since the team benefits and you guys also help to motivate other team members with this. Keep pushing each other, everyone wins in the end.
|
|
|
Post by MachineLearning on Apr 5, 2022 9:04:43 GMT -5
Hell yeah!! I'm competitive, but I'm happy as long as the team is on top. It's far more motivating than just playing for myself. Plus I'm becoming confident in my own ability, the numbers are meaning less. On topic, I bet you can totally nail sub-30, if you can manage to get a C2D close to 5GHz... Now we got #1, #2, #4! I'll be trying again (with safe voltages ) sometime soon..... THIS. Good to see the sprit of competition driving guys to succeed. I mean it's all good in the end since the team benefits and you guys also help to motivate other team members with this. Keep pushing each other, everyone wins in the end. Totally. I think playing just for yourself is a path to unhappiness wih OC. Inevitably your scores will be beaten, and you will drop in the rankings, that's just how any competition works. If you play for your team and for fun, that doesn't matter anymore. Plus as a rule, it's cool to see how far tech will go. On that note, Osmium, let's both crank past 30s next time we sub PyPrime DDR2.
|
|
|
Post by osmiumoc on Apr 5, 2022 9:11:25 GMT -5
On that note, Osmium, let's both crank past 30s next time we sub PyPrime DDR2. You bet! Let's hunt some weak DDR3 scores on our DDR2
|
|
|
Post by Macsbeach98 on Apr 5, 2022 19:35:56 GMT -5
I play to win. Second is only the winner of the losers Its good to see ya's having fun.
|
|
|
Post by Vinster on Apr 5, 2022 20:34:56 GMT -5
|
|
|
Post by osmiumoc on Apr 9, 2022 12:36:03 GMT -5
I welcome DDR2 to the sub 30 second range in PYPrime: hwbot.org/submission/4970812_osmiumoc_pyprime___2b_with_benchmate_ddr2_sdram_29sec_998msAs you can see by the very close result, this was actually not that easy to achieve. I think X48 boards will have a big advantage, as with other DDR2 memory benchmarks, due to the better performance level possible. I was very on the edge of PL9 with any pull-ins resulting in instant death beyond 550 FSB. But in contrast to SuperPI you can't just grab an E8600 and run higher multi with lower FSB for better PL, because you need the high RAM speed. Bandwidth seems very important to PYPrime, I had no luck with 560 1:1 CL4-4-4-4, always a lot slower for me. Man this was fun, finally hitting that sub 30 felt great. And everything was 100% on edge, CPU hanging on for dear life on ambient cooling, RAM was @2.5V (still using some average sticks before abuse my good ones ). I can't wait to see what can be done on DDR2 still with X48 and sub-zero!
|
|
|
Post by MachineLearning on Apr 9, 2022 18:01:43 GMT -5
I welcome DDR2 to the sub 30 second range in PYPrime: hwbot.org/submission/4970812_osmiumoc_pyprime___2b_with_benchmate_ddr2_sdram_29sec_998msAs you can see by the very close result, this was actually not that easy to achieve. I think X48 boards will have a big advantage, as with other DDR2 memory benchmarks, due to the better performance level possible. I was very on the edge of PL9 with any pull-ins resulting in instant death beyond 550 FSB. But in contrast to SuperPI you can't just grab an E8600 and run higher multi with lower FSB for better PL, because you need the high RAM speed. Bandwidth seems very important to PYPrime, I had no luck with 560 1:1 CL4-4-4-4, always a lot slower for me. Man this was fun, finally hitting that sub 30 felt great. And everything was 100% on edge, CPU hanging on for dear life on ambient cooling, RAM was @2.5V (still using some average sticks before abuse my good ones ). I can't wait to see what can be done on DDR2 still with X48 and sub-zero! Great work!! I don't think I can beat this score until I either get some watercooling setup for 775 or better CPUs. None of my E8500's will hit > 4.6GHz stable on air, X5470 maxxes out at 4.3GHz. This benchmark really is fun, I'm happy to see your improvement. I will be doing some Yorkfield vs Wolfdale testing because I want to find out if C2Q have an inherent advantage or disadvantage at the same clock speed in this bench.
|
|
|
Post by osmiumoc on Apr 9, 2022 18:18:22 GMT -5
[...]I will be doing some Yorkfield vs Wolfdale testing because I want to find out if C2Q have an inherent advantage or disadvantage at the same clock speed in this bench. It is a single thread workload from what I saw. I would expect C2Qs to perform slightly worse if the thread jumps from core to core. With fixed affinity on a single core I would expect equal performance. But that is only my theory, would be great to back this up with some testing.
|
|
|
Post by MachineLearning on Apr 9, 2022 18:59:23 GMT -5
[...]I will be doing some Yorkfield vs Wolfdale testing because I want to find out if C2Q have an inherent advantage or disadvantage at the same clock speed in this bench. It is a single thread workload from what I saw. I would expect C2Qs to perform slightly worse if the thread jumps from core to core. With fixed affinity on a single core I would expect equal performance. But that is only my theory, would be great to back this up with some testing. I agree. I assume that there would be more latency due to the cores communicating through the chipset. We shall know shortly
|
|
|
Post by Mr.Scott on Apr 10, 2022 9:04:03 GMT -5
I get better scores using all threads.
|
|