May 12, 201214 yr This is worse that waiting to see what is behind curtain #1! Come on Max, take it easy on us, get with it!!
May 12, 201214 yr Sorry, was typing up a response and had a BSOD because I still had C3 and C6 enabled. Things look solid now! Temps under IBT load testing will cause thermal throttling to kick in, as 105 degrees is reached on occasion. During OCCT testing this gets up to about 92. FSX hits 82 on the hottest core at 4.7GHz during an extended FSXMark 11 run, the other cores peak between 76 and 81. Initial FPS results: Frames, Time (ms), Min, Max, Avg 14218, 300000, 27, 66, 47 Keep in mind this is with no GPU overclocking at all. That is next. I expect results to be near identical to those of 5+ GHz Sandy Bridge systems. I can say with full confidence that the sim is very smooth and pleasing to the eye. I will start GPU overclocking and post results as soon as I have them. Quick question: for the purposes of FSXMark 11 testing, what Affinity Mask is to be used for a 4c/8t CPU? Mine is set to 84 currently, which equates to using all 8 threads if I'm not mistaken. Isn't that less than optimal for FSX? HT is enabled, btw.
May 12, 201214 yr Mine is set to 84 currently That's correct. With 84 you get the same effect than with 14 + HT off, 1 thread per physical core (1, 2 & 3), leaving core 0 for fibers. Can you please confirm PCIe 3.0 is working?
May 12, 201214 yr Max, why are you overclocking for FSX with hyperthreading on? Turn it off to lower the temperature considerably. And set AM to 14.
May 12, 201214 yr I can boot at 4.9 and run IBT for a bit but it's not stable by any means, just too dang hot. I am trying again at 4.7 with HT off and seeking lower temps, working *much* better now. Still peaks at 105 but only on one core and Linpack GFLOPs rating has gone from ~90 to 105+! Turbo speeds to 4.7 are able to be maintained for much longer with these settings, should always be at 4.7 in FSX now. I'm going to stick w/4.7 HT off and keep trying for lower temps.
May 13, 201214 yr 105 ºC? :Shocked: Yep. Thank Intel for their decision to use thermal paste between the core and the heat spreader, rather than fluxless solder as was the case with Sandy Bridge chips. The core heats up and the heat has nowhere to go because it can't be transferred away fast enough. I can touch any part of my cooling system while running IBT and feel virtually zero heat above ambient temperature, that proves the matter to me. Temps are coming down now though. FSX temps are in the 70's and I'm still playing with voltages to get them down further.
May 13, 201214 yr That's Linpack GFLOPs, not temperature. Just so we're clear. Temps are now safely in the 70's under load except for the most extreme stress tests like IBT where it *briefly* peaks at 105 on one core and then throttles back down to 4.0-4.5GHz for a moment until it can crank back up to 4.7. Scores in Linpack are now between 110-116 GFLOPs. Just for reference, I have a dual 8 core Xeon system at work that only scores about 120 GFLOPs so I'm happy with that result :Nerd:
May 13, 201214 yr That's Linpack GFLOPs, not temperature. Just so we're clear. Temps are now safely in the 70's under load except for the most extreme stress tests like IBT where it *briefly* peaks at 105 on one core and then throttles back down to 4.0-4.5GHz for a moment until it can crank back up to 4.7. Scores in Linpack are now between 110-116 GFLOPs. Just for reference, I have a dual 8 core Xeon system at work that only scores about 120 GFLOPs so I'm happy with that result :Nerd: Ahh, ok, so it's the AVX linpack. No wonder it gets so hot! I bet in Prime 95 your temps are much lower, maybe in the mid 80's or something?
May 13, 201214 yr Ahh, ok, so it's the AVX linpack. No wonder it gets so hot! I bet in Prime 95 your temps are much lower, maybe in the mid 80's or something? Maybe IB isn't quite as bad as we think.... The super fast RAM seems to make up for the lesser OC head room.
May 13, 201214 yr I'll try something else and let you know. Looks like a 3rd party site has finally confirmed what has been suspected for several weeks now WRT runaway temps on Ivy: http://vr-zone.com/articles/ivy-bridge-proven-to-suffer-from-poor-thermal-grease/15844.html Very disappointed in Intel right now. This chip is fantastic. Their penny pinching methods are hurting their image. I really hope they come to their senses and fix this. I also hope they offer replacement chips to any early adopters ;)
May 13, 201214 yr Maybe IB isn't quite as bad as we think.... The super fast RAM seems to make up for the lesser OC head room. Yeah, with proper cooling and the little extra IPC, a 4.5GHz IB should perform like a 4.7-4.8GHz SB. PCIe 3.0 seems not to be of much benefit FPS wise though
May 13, 201214 yr Maybe IB isn't quite as bad as we think.... The super fast RAM seems to make up for the lesser OC head room. I was surprised by the measurable performance difference with faster RAM as well, Ben. Now I'm on to PCI-e 2 vs. 3 testing.
Create an account or sign in to comment