This is all well and good but what I'm really interested in is how you got your Haswell @ 4.5-4.6 Ghz to be 30C cooler than IB at the same frequency. I'm sure the OP is very curious about this too...
Nah, I was saying that many users seem to be running with un-optimized voltages and that's how temps on Haswell can be reduced. I wasn't referring to IVB vs. Haswell...as I said in that post. but more on that in a second...
You need to realise that while the first PCIe slot of a Z87 based board will work as a PCIe 3.0 x16, the second slot will only work as a PCIe 2.0 x8. This is way more than a negligible difference and 2 years from now a dual GPU upgrade could cause a bottleneck. There are no GPU cards on the market today that require PCIe 3.0, yet ALL current GPUs will benefit from a x16 PCIe 2.0. Also the OP mentioned that he wanted to use (and improve) this system for the next 3/4 years and hence my comment about this platform not being ideal for SLI/CFire...
I have dual 7950s, dual 7970's, and dual GTX670's. I cannot say I see any detriment to running on Intel with SLI/Crossfire...any loses from PCIe are gained back, and then some via Intel's performance difference. Do keep in mind, I am usually one of the more vocal users about the importance of PCIe link width. Just at that level, there is nothing to be reported. Three and four VGAs...that's a different story.
I didn't accuse you of anything. I do realize however that we're all just people and at times it can be hard for us to be objective.
Over the years I've witnessed a lot of urban legends in the TPU threads regarding magic chips and overclocking feats that were lets just say "exaggerated". In fact I'm still waiting for a certain forum member to show me his Q6600 (C0) running @4.8 Ghz on air and 100% stable... it's been over 5 years now!
Nah, I was just merely saying, since there are many accusations thrown around, that I'm just reporting my own experience.. which may differ from other experiences. But that's part of my job, I think, as a reviewer.
Q6600 @ 4.8 on air...must been a golden chip, water, maybe. possible? Yes. 24/7 stable, probably not. XD
Speaking of "golden" chips, you could also say that my chips are really good, but at the same time, you can check any of our board reviews or memory reviews in the past couple of months to see my results with Haswell. I have no issues with 4.6 GHz, which is what I get out of past Intel chips...since SB. You can go back through all of my board reviews to SKT1156 and see that 4.6 GHz clock used since then. To me, all current Intel chips are capable of that.
What I do see often, at the same time, is users trying to push cache speeds at same multi as CPU, and while that's great for benchmarks, it's part of the heat problem too, IMHO.
I guess, in the end, I don't think anyone should be surprised by the heat output of Haswell. This is what frustrates me the most...since for me, it doesn't affect my OCs. I am at LOWER temps with my Haswell chips vs my IVB chips. Our cooler reviewer is as well. Both of us had retail IVB and Haswell chips. IVB for Crazy was 90C loaded, Haswell..around 80. I am running 4.6 GHz on my Haswell @ 78C loaded, full AVX2. My IVB board chip was around 85C. There's no "Haswell is hotter" correlation...at least not for me. I had over 30 IVB chips, and just 13 Haswells so far, so my sampling is still limited. That's my own experience, across many boards now. Differences in silicon quality are sure to have an impact on people's opinions about this subject. At the same time, few users will have hardware like this:
Nor will they keep track of results like this:
I just do this enthusiast thing differently than most.
I am the SINGLE reviewer that reports power consumption via the 8-pin, and not full system power consumption. With that in mind, Haswell pulls more power than IVB/SNB..so it SHOULD run hotter. My issues isn't with reports of hotter temps, which I feel can be lowered...it's with the reporting that those higher temps cause problems, because that's not the problem for me.
// 631
