|
Cygni posted:I don't think there's gonna be a sudden explosion in CPU parallelism. People predicted that 8 years ago (hence Bulldozer), and it never happened. At the risk of being a parallelism-apologist, I think it's going to happen. The trouble with Bulldozer was that the IPC sucked, was worse than Intel's chips at 4 cores, and so everyone just regarded it as bad and ignored it. The trouble with parallelizing game engines is that it has to seem worth it to the developers, and 4 cores is a god drat rut. Parallel algorithms usually come with higher overhead (except the "embarrassingly parallel" ones). So your complex single thread algorithm naively parallelized gives you 160% of the performance when using all 4 cores. Which sounds like poo poo, so efforts are usually abandoned at that point. But at 8 cores it'd be 320% of the performance, and if they then decide to put the effort in to improve it, they might be able to achieve 6-700% the performance eventually. I'll grant that game engines are probably the most challenging thing to parallelize: you've got 16ms deadlines at 60fps. There's really hard upper limits to the communication costs you can suffer here. But we've only just got parallel rendering APIs like Vulkan/DX12 a year ago. We're only just getting good mainstream CPUs with more than 4 cores this year. We're starting to see some game engines that are proving you can actually use that power. I think the logjam will break.
|
#
?
Jul 19, 2017 16:37
|
|
|
|
| # ? Apr 27, 2024 22:49 |
|
|
I'm glad about the i5's being 6 core. I wonder what will happen to the price.
|
|
#
?
Jul 19, 2017 17:10
|
|
|
~rumors on the anandtech forums~ are that the i5 models will have much more competitive turbo clocks than they used to, the 8700K will still be the fastest processor but it won't be as big of a gap as with the 7700K vs 7600K. I guess they're looking to make the 8600K the "sweet spot" for gamers so it compares favourably to the 6C Ryzen equivalents. eames fucked around with this message at 17:22 on Jul 19, 2017 |
|
#
?
Jul 19, 2017 17:19
|
|
|
B-Mac posted:Anyone have experience delidding a 6700k? I have one core that gets 13C hotter than the others when playing some games, up to 86C spikes at times. Fairly certain the cryorig h7 is mounted properly with max-4 paste and I have no desire to go water. My idles temps are between 25-30C. There's quite a few people on HardOCP with good results for delidding that gen of processor https://hardforum.com/threads/6700k-delid-before-and-after-results.1878870/ Both temp and fan speed way down under the same load on a heavy overclock. Keep in mind that most games favor one of their cores more heavily than others and you will see a temp differential because that one will be hitting turbo clocks more than the rest. But it does seem like a good candidate regardless.
|
|
#
?
Jul 19, 2017 19:30
|
|
|
Outside of the Intel NUC, is there no way to obtain a CPU with an iris pro 580? I'd really love to build something with an 6685R.
|
|
#
?
Jul 20, 2017 03:24
|
|
|
I've been using a 4 core/8 thread sandy bridge laptop since I got it in 2011. I credit a lot of how it holds up, despite the normal clock speed being 2.0 ghz, to simply having all those threads around. It's because of this that I'd refuse to get a plain 4 core CPU for anything meant to be used a lot now, especially with the way browsers love multiprocess now. Maybe a 6c/6t could be acceptable because of having 6 real cores trumping 8 logical cores, but I'd say 4c/4t is a foolish thing to buy new.
|
|
#
?
Jul 20, 2017 03:33
|
|
|
Xeom posted:Outside of the Intel NUC, is there no way to obtain a CPU with an iris pro 580? I'd really love to build something with an 6685R. You can get a Thinkpad P70 with the mobile Xeon option, but it'll cost you as much as a used car. The desktop models listed on Ark seem to be literally nonexistent, I don't know what uses them if anything. Eletriarnation fucked around with this message at 03:48 on Jul 20, 2017 |
|
#
?
Jul 20, 2017 03:46
|
|
|
Bought the rockitcool delid and relid kit along with some thermal grizzly condocunaut. Full yolo hope I don't destroy my CPU.
|
|
#
?
Jul 20, 2017 21:41
|
|
|
fishmech posted:I've been using a 4 core/8 thread sandy bridge laptop since I got it in 2011. I credit a lot of how it holds up, despite the normal clock speed being 2.0 ghz, to simply having all those threads around. 6C/6T seems fine, I can't think of any situation ever where 4C/8T would be preferable to 6C/6T.
|
|
#
?
Jul 20, 2017 21:50
|
|
|
Twerk from Home posted:6C/6T seems fine, I can't think of any situation ever where 4C/8T would be preferable to 6C/6T. Price (die size), power consumption, and presumably the potential for slightly higher clocks and thus single-thread performance. Not that I disagree with you; for most cases, 6/6 is a better option than 4/8.
|
|
#
?
Jul 20, 2017 21:53
|
|
|
meanwhile the PC Building/Upgrading/Parts-picking megathread still recommends "Four. Maybe two." 
|
|
#
?
Jul 20, 2017 21:55
|
|
|
DrDork posted:Price (die size), power consumption, and presumably the potential for slightly higher clocks and thus single-thread performance. Remember, in most games the hyperthread is a ~10% improvement in performance tops. The real-world maximum is about 30% but you don't typically hit that in games. This is actually the i5 that I was hoping Intel would put on the X299 platform (ideally based on SKL-X so you get the quad-channel RAM - but I guess SKL-X hasn't panned out super well so far)
|
|
#
?
Jul 20, 2017 22:02
|
|
|
eames posted:meanwhile the PC Building/Upgrading/Parts-picking megathread still recommends "Four. Maybe two." Are they wrong? If you're building a budget gaming PC at this instant, and going down to a 2C/4T CPU gets you to a GTX 1060 and a decent SSD that's entirely worth it, especially given how much Intel charges for a basic quad core.
|
|
#
?
Jul 20, 2017 22:06
|
|
|
Twerk from Home posted:Are they wrong? If you're building a budget gaming PC at this instant, and going down to a 2C/4T CPU gets you to a GTX 1060 and a decent SSD that's entirely worth it, especially given how much Intel charges for a basic quad core. Yeah, Intel is the key word here. Two cores only makes sense at the absolute rock bottom (G4560) or office PCs, anything above that is Ryzen territory. More cores and a free upgrade path. Ryzen 3 will pretty much be the end of the dual core era and 2C/4T will probably bottleneck a 1060 within a year or two.
|
|
#
?
Jul 20, 2017 22:17
|
|
|
Twerk from Home posted:Are they wrong? If you're building a budget gaming PC at this instant, and going down to a 2C/4T CPU gets you to a GTX 1060 and a decent SSD that's entirely worth it, especially given how much Intel charges for a basic quad core. "Power users" always forget that regular humans exist, who dont want to spend a lot of money on the thing they use to read twitter, do their taxes, and occasionally play overwatch on at 1080p. Heres a 3.9Ghz 2C/2T i3 vs a 3.9Ghz (all core turbo) 4C/4T i5: http://www.anandtech.com/bench/product/1889?vs=1865 Sure its faster in most games, but the i3 is nearly exactly half the price. And when paired with a mid tier GPU like a 1060, or lower like a 1050 Ti, you're GPU limited in nearly every game anyway...
|
|
#
?
Jul 20, 2017 22:22
|
|
|
Paul MaudDib posted:Remember, in most games the hyperthread is a ~10% improvement in performance tops. The real-world maximum is about 30% but you don't typically hit that in games. Sure, I'm just saying that if you can make a 3.8GHz 6C, you could probably have made it a 4.1GHz or similar 4C, and there are some applications where that might be a worthwhile tradeoff because whatever you're doing happens to be single-thread limited. But in general more cores is more better, and I too am glad to see Intel finally dipping their toes into the 6C consumer marketplace.
|
|
#
?
Jul 20, 2017 22:40
|
|
|
B-Mac posted:Bought the rockitcool delid and relid kit along with some thermal grizzly condocunaut. Full yolo hope I don't destroy my CPU. I did this for my 4790k, reduced temp by 15c
|
|
#
?
Jul 20, 2017 23:33
|
|
|
Cygni posted:"Power users" always forget that regular humans exist, who dont want to spend a lot of money on the thing they use to read twitter, do their taxes, and occasionally play overwatch on at 1080p. My parents' PC is pretty much just a print server at this point, as that is the reason for its existence in 98% of use cases.
|
|
#
?
Jul 21, 2017 00:06
|
|
|
Visidan posted:I did this for my 4790k, reduced temp by 15c Hope I get results that great.
|
|
#
?
Jul 21, 2017 00:20
|
|
|
Cygni posted:"Power users" always forget that regular humans exist, who dont want to spend a lot of money on the thing they use to read twitter, do their taxes, and occasionally play overwatch on at 1080p. Which wants more cores than 2, because it's the year 2017, and websites are total hogs, especially normal people sites like Facebook.
|
|
#
?
Jul 21, 2017 00:58
|
|
|
eames posted:meanwhile the PC Building/Upgrading/Parts-picking megathread still recommends "Four. Maybe two." there really is nothing better than the g4560 in that price range though
|
|
#
?
Jul 21, 2017 01:37
|
|
|
Anime Schoolgirl posted:honestly it's pretty disappointing to see that the r5 1600 isn't on the recommended list and the i5-7600k still is aren't those megathreads only updated every 4 months for the first 3 years and then not updated for 3 more years?
|
|
#
?
Jul 21, 2017 01:41
|
|
|
it's more "updated once, ever and never again" now
|
|
#
?
Jul 21, 2017 01:43
|
|
|
fishmech posted:Which wants more cores than 2, because it's the year 2017, and websites are total hogs, especially normal people sites like Facebook. The future of computing: Imagine ads and data tracking to the face, forever
|
|
#
?
Jul 21, 2017 02:09
|
|
|
I'm convinced buying a 2600K at launch was the best CPU purchase I've ever made. I think my laptop is also Sandy Bridge, and my old MBP is a Nehalem; unfortunately, that one limps a little bit with websites that are full of lovely JavaScript. Or that's just an old OS X install and Safari sucking...
|
|
#
?
Jul 21, 2017 03:03
|
|
|
This is probably a stupid question, but can somebody explain in layman terms why more cores almost always means less speed?
|
|
#
?
Jul 21, 2017 06:46
|
|
|
It's a combination of a few factors. Power draw and heat output are both big factors, but a lot of it comes down to the fact that they have to coordinate together, and the more things need to be kept in sync the harder it is to do that.
|
|
#
?
Jul 21, 2017 06:48
|
|
|
TorakFade posted:This is probably a stupid question, but can somebody explain in layman terms why more cores almost always means less speed? Moar cores on one die also mean moar chance of one not to hit a rated speed. So they get better yields by running them at a lower clock also. Also without process efficiency changes, if you have a fixed power target, less cores can clock higher within the same target. GRINDCORE MEGGIDO fucked around with this message at 16:05 on Jul 21, 2017 |
|
#
?
Jul 21, 2017 07:09
|
|
|
TorakFade posted:This is probably a stupid question, but can somebody explain in layman terms why more cores almost always means less speed? Physical factors as cited are one good reason, but another reason is, as Sinestro said, that you still cannot parallelize works without introducing any kind of overhead. That overhead grows more as you introduce more worker cores. https://home.wlu.edu/~whaleyt/classes/parallel/topics/amdahl.html
|
|
#
?
Jul 21, 2017 08:13
|
|
|
Your weekly dose of "lol X299" https://twitter.com/HardwareUnboxed/status/888225519141093376
|
|
#
?
Jul 21, 2017 11:05
|
|
|
X299 looking great. Good work Intel.
|
|
#
?
Jul 21, 2017 16:07
|
|
|
Wow there has been some rumors that Intel was going to do redo the socket somehow for X299 due to heat/power issues but I didn't think they were legit at all. Normally BS problems like that doesn't get into the finished for sale product.
|
|
#
?
Jul 21, 2017 17:22
|
|
|
I feel like that's a faulty socket. The pad and pin should be able to handle more current than the traces on the CPU, but if one of the pins wasn't properly springy due to a manufacturing fault or a bit of dust bent it down so it had a weak connection then it'd fail. Skylake-X has all sorts of other problems, but I don't think this specifically is a reason not to get it if somehow the other problems aren't a big deal for you.
|
|
#
?
Jul 21, 2017 17:30
|
|
|
Maybe it was a contact issue and sparks flew? The trace is probably way thinner than the contact surface of the pin and pad, and it didn't fry apparently.
|
|
#
?
Jul 21, 2017 18:37
|
|
|
I do believe it's just too much current for the socket when the CPU is overclocked. The issue isn't completely new, people had similar problems with Socket 1156 at "only" 160W. http://www.anandtech.com/show/2859
|
|
#
?
Jul 21, 2017 18:39
|
|
|
That article blames Foxconn for lovely sockets, though, coincidentally with bad pin to pad contact.
|
|
#
?
Jul 21, 2017 18:46
|
|
|
More Coffee Lake leaks, launching August 22-26: http://wccftech.com/intel-coffee-lake-core-i7-8700k-6-core-cpu-specifications-details-leak/ 3.7 base, 4.3 single core, 4.0 all core turbo on the 8700k. Doesn't mention a new revision of 1151, either. That locked 65w model looks pretty tasty too, although I cant tell if hit has HT. 7700k for comparison is 4.2 base, 4.5 single core, 4.4 all core, so you give a couple hundred mhz back but gain 4 threads.
|
|
#
?
Jul 22, 2017 00:09
|
|
|
Wait, they're doing Z390? Is that supposed to be higher end than Z370 or an outright replacement?
|
|
#
?
Jul 22, 2017 13:01
|
|
|
Yup, I can definitely wait one more month for Coffee Lake especially if I can reuse my Z170 mobo.
|
|
#
?
Jul 22, 2017 13:25
|
|
|
|
| # ? Apr 27, 2024 22:49 |
|
|
Rabid Snake posted:Yup, I can definitely wait one more month for Coffee Lake especially if I can reuse my Z170 mobo. That will probably depend on a bios update. I've got a Gigabyte Z170 and am hoping they update the bios to support coffee lake. Hey, here's a question, if I change CPUs do I need to reinstall windows? What if I change motherboards to (say, to a Z270)? Question pertains both to weird driver issues, and also Windows activation.
|
|
#
?
Jul 22, 2017 15:23
|
|
