|
Lowen SoDium posted:I agree with all of this. But I do wonder about the TIM thing. It's been established that the problem is not the TIM specifically, but more the spacing because of the glue, right? Is there any chance that that has been corrected or will be better on these chips than it has been in the past?
|
# ? Jun 6, 2017 21:40 |
|
|
# ? May 20, 2024 04:48 |
|
--edit: Actually, nevermind, those Wccftech numbers don't make much sense. Goddamnit.
Combat Pretzel fucked around with this message at 22:07 on Jun 6, 2017 |
# ? Jun 6, 2017 22:03 |
|
My 3570K doesn't consistently boot beyond 4.2GHz so I don't even need to worry about delidding for temps.
|
# ? Jun 6, 2017 22:11 |
|
GRINDCORE MEGGIDO posted:Maybe, but they could have improved it at any point after Ivy, and they didn't. This seems to be the crux of a lot of the complaints at this point. It's not necessarily that these chips and X299 are bad as much as they give off a pretty strong impression that Intel just doesn't give a poo poo about customers and is attempting to nickle and dime you and/or do the least amount possible when it would be trivial for them to provide a more compelling product. Turns out people--as a general rule--don't appreciate that, even if, were those trivial improvements unavailable entirely, they'd otherwise be happy with the product. Just knowing that it could be so much better makes people annoyed.
|
# ? Jun 6, 2017 22:52 |
|
I might have a very niche use for 8 cores, but right now my 6 core Haswell does 4.7GHz. I'd be forced to make a hard decision if 8 core Skylakes ended up not doing at least 4.7GHz on average. Even "only" doing 4.7GHz wouldn't feel great because 99% of the stuff I do likes single thread performance more than lots of cores so I'd be spending probably nearly 1,000 dollars for nearly a lateral upgrade.
|
# ? Jun 6, 2017 22:58 |
|
Intel thermal interface mayo
|
# ? Jun 6, 2017 23:08 |
|
craig588 posted:I might have a very niche use for 8 cores, but right now my 6 core Haswell does 4.7GHz. I'd be forced to make a hard decision if 8 core Skylakes ended up not doing at least 4.7GHz on average. Even "only" doing 4.7GHz wouldn't feel great because 99% of the stuff I do likes single thread performance more than lots of cores so I'd be spending probably nearly 1,000 dollars for nearly a lateral upgrade. Using Haswell-E as a baseline, you are probably looking at more like 9% IPC improvement and 7% clockrate improvement. But yeah, it's not an earthshattering difference given the fact that Haswell-E has been out for 3 years now. Paul MaudDib fucked around with this message at 23:12 on Jun 6, 2017 |
# ? Jun 6, 2017 23:10 |
|
Alpha Man posted:Intel thermal interface mayo This is NOT how you use thermal paste. Or mayo.
|
# ? Jun 8, 2017 00:38 |
|
mewse posted:This is NOT how you use thermal paste. Or mayo. Just a single pea sized drop right in the middle for either
|
# ? Jun 8, 2017 05:46 |
|
mewse posted:This is NOT how you use thermal paste. Or mayo. Well yeah, even I know that for best results you want to be using chocolate.
|
# ? Jun 8, 2017 06:03 |
|
Actually you're supposed to use the mayo but you need a knife to really press the mayo into all the nooks n' crannies of the socket or it won't work right. Chocolate is just the latest fad.
|
# ? Jun 8, 2017 08:02 |
|
2013 base Crystalwell rMBP vs the new 2017 base Kaby Lake MBP, 4 years of Intel mobile CPU progress! (I assume it's the 128MB eDRAM L4 cache). http://browser.geekbench.com/v4/cpu/compare/3038252?baseline=3050865
|
# ? Jun 8, 2017 09:35 |
yeah it seems so. everything that the 2013 one is better at seems to either be explicitly or subtly reliant on cache like there are explicit memory tests, and id not be surprised if dijkstra and speech recognition loves access to enough memory to store everything in Watermelon Daiquiri fucked around with this message at 10:27 on Jun 8, 2017 |
|
# ? Jun 8, 2017 10:25 |
|
eames posted:2013 base Crystalwell rMBP vs the new 2017 base Kaby Lake MBP, 4 years of Intel mobile CPU progress! (I assume it's the 128MB eDRAM L4 cache). Did Apple stop ordering the SKUs that had 64MB of eDRAM L4 cache on them? The 6260U and friends sure did.
|
# ? Jun 8, 2017 15:35 |
|
IIRC Skylake GT3e/GT4e was delayed for a long time so Apple put Polaris 11 into every 15". That eliminated the need for a powerful iGPU, so they switched to the cheaper SKUs without eDRAM. The fact that Intel's iGPU roadmap was starting to look a bit bleak compared to Nvidia/AMD probably also had something with it. The smaller 13" models probably still have eDRAM because there's no room for a dGPU.
|
# ? Jun 8, 2017 15:51 |
|
There's a PCI-SIG conference going on now, and some announcements. It looks like PCIe 4 is now all-but-published: http://techreport.com/news/32064/pcie-4-0-specification-finally-out-with-16-gt-s-on-tap Just waiting on the lawyers for the spec to be final. Still no word from AMD or Intel about a timeline for support. SMI has a roadmap with PCIe 4 SSD controllers at end of next year: https://tweakers.net/ext/f/tv4gcfsSQVuoif5T3coYAc28/full.png PCIe 5 announced, and they're actually shooting for 32 Gbps per lane, which is awesome. Also, apparently a super-aggressive 2019 spec target date, though at this point they're famous for delays... https://www.techpowerup.com/234167/pci-sig-fast-tracks-evolution-to-32-gt-s-with-pci-express-5-0-architecture
|
# ? Jun 8, 2017 16:58 |
|
Josh Lyman posted:My 3570K doesn't consistently boot beyond 4.2GHz so I don't even need to worry about delidding for temps. Same but with my 2500K. It just won't go above 4.2GHz and be stable no matter what I do and it's infuriating. And now that Coffee Lake has been pushed back to next year I'm probably going to go with a R5 1600X or an i7-7700K build this fall. 6 cores/12 threads or 4 faster cores/8 faster threads?
|
# ? Jun 8, 2017 19:00 |
|
crazypenguin posted:There's a PCI-SIG conference going on now, and some announcements. It looks like PCIe 4 is now all-but-published: The IBM POWER9 has PCIe Gen 4 slots apparently, but good luck getting one of those if you're not a massive hyperscale or enterprise customer. One thing that was being reported incorrectly about gen4 is that the slot provides 300W now, which is wrong. There is an additional spec for a 2x4 power connector to bump total power to 300W. So it'll be pretty similar to the videocard power cables from before. -up to 75W - slot power -up to 150W - slot & previously existing 2x3 power (+75W) -up to 300W - slot & 2x3 & additional 2x4 connector (+150W) I thought that was interesting, and had seen tech sites reporting the wrong info that it all came off the slot.
|
# ? Jun 8, 2017 19:28 |
|
priznat posted:The IBM POWER9 has PCIe Gen 4 slots apparently, but good luck getting one of those if you're not a massive hyperscale or enterprise customer. PCIe 3 tops out at what, 30W per slot without additional connectors? Should be nice for some of the absurdly dense NVMe devices that I assume are coming up.
|
# ? Jun 8, 2017 21:18 |
|
BangersInMyKnickers posted:PCIe 3 tops out at what, 30W per slot without additional connectors? Should be nice for some of the absurdly dense NVMe devices that I assume are coming up. 75W, you can get GTX 1050 Tis that run without a 6-pin power connector I believe. 1050s for sure.
|
# ? Jun 8, 2017 21:26 |
|
BangersInMyKnickers posted:PCIe 3 tops out at what, 30W per slot without additional connectors? Should be nice for some of the absurdly dense NVMe devices that I assume are coming up. 75W for anything larger than a x4, otherwise 25W. There's also a 225W spec now, which is slot power & 2x4 150W. These are all total power numbers, 3.3 & 12v. Only the slot provides 3.3 not the aux connectors. I think the high power ones will be for absurdly high power draw GPU and offload cards. The NVMe u.2 ff is probably limited to 25W but I'm not 100% on that..
|
# ? Jun 8, 2017 21:30 |
|
https://www.youtube.com/watch?v=cnhXIL_E_Vc&t=208s "that thing's gonna cook" alright
|
# ? Jun 9, 2017 16:38 |
|
"Go buy an 18core i9 if you really want to Game" Go gently caress yourself. Even the host couldn't hide his revulsion.
|
# ? Jun 9, 2017 17:12 |
|
BangersInMyKnickers posted:"Go buy an 18core i9 if you really want to Game" Not enough EXXXtreme decals or red paint on it to be a Gamer chip. Intel needs to learn from Fatal1ty about gaming hardware
|
# ? Jun 9, 2017 18:58 |
|
DrDork posted:Not enough EXXXtreme decals or red paint on it to be a Gamer chip. Intel needs to learn from Fatal1ty about gaming hardware Hey that had that skull canyon NUC with a skull on it!
|
# ? Jun 9, 2017 20:41 |
|
BangersInMyKnickers posted:"Go buy an 18core i9 if you really want to Game" I guess in one Intel slide at Computex they were touting the i9 for "12k gaming." I mean come the gently caress on you can't even do 8k properly with SLI 1080 Tis, even 5k is pushing the limits of practicality in modern games. MaxxBot fucked around with this message at 05:41 on Jun 10, 2017 |
# ? Jun 10, 2017 05:38 |
|
If they'd somehow made it so an i9 turbo clocked to 5Ghz on four cores for gaming loads, that might be a nice checkmark in its favor. As it stands, I wouldn't be surprised if we don't see a "Kaby Lake-XE" with eDRAM, just introduced too late for anyone to give a drat. God knows the package has the room for it. =/ BIG HEADLINE fucked around with this message at 05:51 on Jun 10, 2017 |
# ? Jun 10, 2017 05:47 |
|
MaxxBot posted:I guess in one Intel slide at Computex they were touting the i9 for "12k gaming." I mean come the gently caress on you can't even do 8k properly with SLI 1080 Tis, even 5k is pushing the limits of practicality in modern games. "12K" as is commonly marketed is an incorrect way of naming 3x4K multi-monitor (11,520x2,160), so it is actually less than "8K" as properly marketed (7,680x4,320). But then again "4K" is actually 4096x2160 not 3840x2160, so whatever, hail satan
|
# ? Jun 10, 2017 05:52 |
|
BurritoJustice posted:"12K" as is commonly marketed is an incorrect way of naming 3x4K multi-monitor (11,520x2,160), so it is actually less than "8K" as properly marketed (7,680x4,320). Why did you think inherently and incontrovertibly quantifiable phenomena would be exempt from 'lol nothing matters'? In relevant news: Brazen intimidation over Microsoft's x86-on-ARM interpolation! Who else actually has anything vaguely resembling a commitment to x86 these days, Intel? Do you really want to tango with Microsoft over this?
|
# ? Jun 10, 2017 06:02 |
|
Yeah just when after all these years we might get a full desktop enabled x86 running phone.. NOPE. COCKBLOCKED. It's ok though MS lost me years ago to Android. Windows just sucks rear end on mobile.
|
# ? Jun 10, 2017 14:45 |
|
https://www.youtube.com/watch?v=kpoies2JcmI
|
# ? Jun 10, 2017 15:51 |
|
Not to be too but this is the guy that visited the Intel HQ at Oregon to talk to them about overclocking partnerships and what not. He also designs, produces and sells SL-X delid tools. Him admitting to use a pretested CPU means it is likely one of the top 5%. Impressive clockspeeds regardless, looks like KL vs Ryzen continued in HEDT. PS: 94°C @4.8Ghz/1,25V on water with an open air test bench. Noctua & co have their work cut out for them.
|
# ? Jun 10, 2017 16:28 |
|
Delids weren't exactly required to hit high 4Ghz-ish numbers though with Kabylake. Heat was still terrible but you had a decent shot of pulling it off with a stock CPU IHS. It looks like delids may a requirement to hit high 4Ghz-ish numbers with Skylake-X. Its not terribly hard or expensive to do but it shouldn't really be necessary considering the cost of those chips and I doubt many will do it.
|
# ? Jun 10, 2017 17:05 |
|
MaxxBot posted:I guess in one Intel slide at Computex they were touting the i9 for "12k gaming." I mean come the gently caress on you can't even do 8k properly with SLI 1080 Tis, even 5k is pushing the limits of practicality in modern games. My "ultimate computing device" would be a 60/65 inch display, Retina (220-230dpi) quality, full touch screen - allowing you to have a desktop the size of, well, a desk. I'd need 12288x9216 pixels to do it, basically 16 4k (4 wide by 4 tall)displays stitched together (113 megapixels, imagine driving that at 120hz!) We have a ways to go. Especially if this thing is going to be driven by my phone/tablet, docking it when I get to my desk at work or home. Bob Morales fucked around with this message at 17:55 on Jun 10, 2017 |
# ? Jun 10, 2017 17:53 |
|
redeyes posted:Yeah just when after all these years we might get a full desktop enabled x86 running phone.. NOPE. COCKBLOCKED. It's ok though MS lost me years ago to Android. Windows just sucks rear end on mobile. Uh, are you really dumb enough to think a smartphone-class ARM SOC running x86 emulation was going to be at all acceptable for performance? There are already actual x86 mobile small devices, which run massively better than the emulator solution ever could: http://www.gearbest.com/tablet-pcs/pp_364503.html
|
# ? Jun 10, 2017 17:57 |
|
dont be mean to me posted:Why did you think inherently and incontrovertibly quantifiable phenomena would be exempt from 'lol nothing matters'?
|
# ? Jun 10, 2017 22:31 |
|
Anime Schoolgirl posted:if this causes Surface Book 5 to use small Ryzen+igpu gently caress yes
|
# ? Jun 10, 2017 22:34 |
|
Anime Schoolgirl posted:if this causes Surface Book 5 to use small Ryzen+igpu Vega graphics surface pro 5 I want to believe
|
# ? Jun 10, 2017 22:46 |
|
fishmech posted:Uh, are you really dumb enough to think a smartphone-class ARM SOC running x86 emulation was going to be at all acceptable for performance?
|
# ? Jun 11, 2017 00:37 |
|
|
# ? May 20, 2024 04:48 |
|
fishmech posted:Uh, are you really dumb enough to think a smartphone-class ARM SOC running x86 emulation was going to be at all acceptable for performance? Um. https://www.youtube.com/watch?v=A_GlGglbu1U&t=79s
|
# ? Jun 11, 2017 01:09 |