|
GRINDCORE MEGGIDO posted:Is Coffeelake 6 core gonna be compatible with Z170, any chance? No one knows yet. The Z270 might have a better shot of it. Just figure that the quads will likely work on the Z170/270 with a BIOS update, but the six-cores might be H370/Z370 only. Coffee Lake is just a Skylake optimization process, it's not a new chip, so who knows. I'd like to think two new physical cores would require motherboard traces to the PCH that just won't be there on the 100 and 200-series boards. BIG HEADLINE fucked around with this message at 23:47 on Apr 19, 2017 |
#
?
Apr 19, 2017 23:45
|
|
|
|
| # ? Apr 20, 2024 17:06 |
|
|
I don't really know what the breakdown looks like for CPU pins, but I doubt they actually need more to add more cores. Pins are mostly PCie lanes and DRAM traces right? Now I'm curious. e: http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/4th-gen-core-family-desktop-vol-1-datasheet.pdf okay mostly voltage sink/sources, followed by ram, followed by other poo poo. I guess a socket could be power inadequate for handling higher core counts, without dropping the clock speed or something. crazypenguin fucked around with this message at 00:02 on Apr 20, 2017 |
|
#
?
Apr 19, 2017 23:51
|
|
|
I think they could easily disavow support that way, but it won't be because of hardware incompatibility. e- yea I think it's more likely the quads will work but in TYOOL 2017 who knows. Plus with mobo manuf. like Asus not even implementing Speed Shift on desktop Skylake....
|
|
#
?
Apr 19, 2017 23:53
|
|
|
Palladium posted:I can wait for a looooong time. After my GTX1070 last year nothing really catches my wallet's attention in PC land anymore. Same goes for phones. There are enthusiasts who still happily run OC'd i7-2600k..
|
|
#
?
Apr 20, 2017 00:09
|
|
|
I can actually see a lot of potential for Coffee Lake to flop. Kaby's lead mostly derives from its exceptional clocks. You can number the architectures that could reliably (ish) hit 5.0 GHz on one hand: Pentium 4, Sandy Bridge, Bulldozer, and Kaby are all that's coming to mind. Two of those were utter trash that were designed for all-out clocks and shat the bed on IPC. Maybe there's some SPARCs or something too? It wouldn't be the first time Intel's tweaks have dinged clock rates real hard. Ivy Bridge lost a whoooole lot of clocks over Sandy Bridge (what happened there anyway?). Or the recent example - Broadwell, where you had like 5% IPC gains but like a 10% loss in clockrate. Even the loss of 400 MHz would seriously ding Intel's cred in the gaming community with Ryzen snapping at their heels like that. And again, Kaby is an exceptionally high clocking architecture to begin with. Unless it was made up with a corresponding gain in IPC (which seems unlikely for a "tweak"). The project managers at Intel have to be sweating bullets right now 
Paul MaudDib fucked around with this message at 02:04 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 02:02
|
|
|
Paul MaudDib posted:architectures that could reliably (ish) hit 5.0 GHz on one hand: Pentium 4 Naaaaahh, even with LN2 5GHz wasn't a guarantee, let alone stabily. Sandybridge could have it happen on air, but only with a good chip and a good cooler. Bulldozer could have it happen on air, but you still needed a good cooler. Kabylake is the first one where with a random chip and a random cooler you had decent odds of hitting 5GHz.
|
|
#
?
Apr 20, 2017 02:18
|
|
|
Paul MaudDib posted:Maybe there's some SPARCs or something too? POWER8  quote:Ivy Bridge lost a whoooole lot of clocks over Sandy Bridge (what happened there anyway?) New, less mature process and the switch away from soldered IHS. quote:Even the loss of 400 MHz would seriously ding Intel's cred in the gaming community with Ryzen snapping at their heels like that. For the "It's useless garbage because it only gets 180 fps instead of 195 at 720p even though the game in question is GPU limited at any settings anyone would actually play it at" crowd, sure. For the rest of us, not so much. I'll gladly pickup a six core Coffee Lake if it works on my Z170 board and I can get ~4.2+GHz out of it and couldn't care less that it would be imperceptibly slower than my 4.5GHz 6700k at single threaded stuff. Otherwise I'll be building a 6 or 8 core Zen+ system when they drop. Theris fucked around with this message at 02:28 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 02:25
|
|
|
Really wonder how they'll price them. I'm absolutely doubting they'll be competitive on price. Just LOL if they lock out the z170 from running 6 cores, too. I hope they don't do that.
|
|
#
?
Apr 20, 2017 02:27
|
|
|
craig588 posted:Naaaaahh, even with LN2 5GHz wasn't a guarantee, let alone stabily. Sandybridge could have it happen on air, but only with a good chip and a good cooler. Bulldozer could have it happen on air, but you still needed a good cooler. Kabylake is the first one where with a random chip and a random cooler you had decent odds of hitting 5GHz. I thought some of the later ones could clock that high (Prescott?). Looking back I guess not. Still pretty good for the time though. Oh well, point remains though, Kaby is an anomaly and I'm curious to see whether Intel will be able to repeat it. Theris posted:POWER8 Knew there had to be something from the RISC contingent. quote:For the "It's useless garbage because it only gets 180 fps instead of 195 at 720p even though the game in question is GPU limited at any settings anyone would actually play it at" crowd, sure. For the rest of us, not so much. I'll gladly pickup a six core Coffee Lake if it works on my Z170 board and I can get ~4.2+GHz out of it and couldn't care less that it would be imperceptibly slower than my 4.5GHz 6700k at single threaded stuff. Otherwise I'll be building a 6 or 8 core Zen+ system when they drop. Well... the caveat is always "at a given price". If it ends up being a Broadwell-style flop where total performance goes down in exchange for a moderate improvement in perf/watt... it becomes a lot harder to make an argument for premium pricing in the desktop market. Because overall the Ryzens are actually the nicer processor at most of the price points. Apart from the 7700K - the entry-level HEDT (5820K/6800K), or the "money is no object" 10C+ market segments are basically the only other places where Intel is competitive right now. And X99 is on its way out right now, stock is dwindling. Right now they are getting spanked in the mid-market segment by the R5s and when the R3s launch the i3s will be pretty much toast too. Since the i3/i5/i7 really represent more of a "good/better/best" at their respective price points than an actual specific feature set, I don't see them lowering their prices, I think they will push better hardware instead. I expect everything to move down one notch. We've already got the hyperthreaded Pentiums, the 4C4T become the i3s, the 4C8T become the i5s, and the 6C12T become the i7s. Potentially with a new crossover in the middle between the consumer and HEDT lineups - 4C8T on the HEDT platform (basically a HEDT i5), plus 6C12T on the consumer platform. As such I really don't see them holding out on the Z370 6C12T, but rather leaning on enthusiasts to go upwards into the HEDT platform with the transitional HEDT i5, and potentially trying to keep the cost of the X299 chipset down to keep the motherboard prices reasonable. Paul MaudDib fucked around with this message at 02:43 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 02:40
|
|
|
GRINDCORE MEGGIDO posted:Really wonder how they'll price them. I'm absolutely doubting they'll be competitive on price. I've a feeling we'll know well in advance since that's the kind of information that while almost certainly under NDA tends to leak out when someone runs a benchmark or posts a CPU-Z shot.
|
|
#
?
Apr 20, 2017 05:32
|
|
|
That _has_ to be a paper launch..... right? I mean surely they won't have silicon boxed up and ready to go as soon as Computex kicks off?
|
|
#
?
Apr 20, 2017 06:32
|
|
|
They won't, of course, CL is launching in August according to that announcement. Computex is only for announcing Sky and Kaby-X.
|
|
#
?
Apr 20, 2017 07:38
|
|
|
Paul MaudDib posted:I can actually see a lot of potential for Coffee Lake to flop. Kaby's lead mostly derives from its exceptional clocks. They seem fairly confident  article posted:14nm++ is claimed to offer 25% greater performance at a given power level than the unoptimized 14nm process first used to produce Broadwell and Skylake chips, or as much as 52% less power consumption for the same level of performance. In fact, Intel's projections show that the transistor performance of 14nm++ will actually exceed that of its first generation of 10-nm products. Expect to see 14nm++ underpin Intel's rumored Coffee Lake CPUs later this year. I wonder if the Basin Falls line (Skylake-X/Kaby Lake-X) is going to use the same process. http://techreport.com/review/31660/intel-defends-its-process-technology-leadership-at-14nm-and-10nm
|
|
#
?
Apr 21, 2017 07:32
|
|
|
Eletriarnation posted:Ah, is that only possible if they're idling though or is non-turbo operation allowed? I was thinking of a scenario where you have a two-threaded main load and lesser threads which can be handled by the other cores, like the OS background processes. I should try selectively loading some Xeons at work with Prime95 or Intel's tuning utility and see what happens, I guess. As far as I know Intel's turbo control algorithm doesn't care if a thread is "lesser". To it, a core that is powered up and running code is active. If 1 core is active, and all the rest are idle, it permits anything up to the highest possible multiplier boost. If 2 are active, it sets the max possible multiplier boost for both cores to the 2-core entry in the table. So on and so forth. There might be some leeway in the evaluation of what's "active" so that it doesn't have too much of a hair trigger in dialing back the max multiplier boost, but everything I've read about it (in Intel's own literature and AnandTech articles and whatnot) says that the limit the power controller select is applied globally across all cores, not individualized per core.
|
|
#
?
Apr 21, 2017 09:10
|
|
|
eames posted:They seem fairly confident
|
|
#
?
Apr 22, 2017 08:04
|
|
|
So will the six-core Coffee Lake chips be the i7 line only? Or will we see six-core i5 chips? I'm hoping we'll see at least a 6C/6T i5 chip for ~$250 if not then I'll probably still build a 1600X rig sometime this summer.
|
|
#
?
Apr 22, 2017 23:43
|
|
|
spasticColon posted:So will the six-core Coffee Lake chips be the i7 line only? Or will we see six-core i5 chips? I'm hoping we'll see at least a 6C/6T i5 chip for ~$250 if not then I'll probably still build a 1600X rig sometime this summer. There's no way of knowing without someone breaking what's probably still a pretty iron-clad NDA. Again, it's just a guess based on logic, but I wouldn't expect the 6C/?T Coffee Lakes until ~6 months *after* Sky/Kaby-X drop. The dual/HTs and quads will likely launch first as to not give their own 'flagship enthusiast' product an attractive competitor. BIG HEADLINE fucked around with this message at 23:53 on Apr 22, 2017 |
|
#
?
Apr 22, 2017 23:46
|
|
|
To be fair, a Skylake-E 6 core and a Coffee-Lake 6 core wouldn't conflict as much as you'd think. LGA2066 gets you quad channel memory and massive amounts of PCI-E lanes while LGA1151 gets you better clocks and an iGPU. They'll probably be roughly the same price as with the 7700K/6800K, ignoring platform cost differences, which lines them up as meaningful sidegrades depending on use case. The platform features are almost all dependant on chipset and I'm sure Intel will line them up so there won't be a big delta like there is with X99/Z270 right now. drat I'm excited for Coffee-Lake to hopefully hit the best of both worlds of the 7700K and Ryzen/HEDT Intels.
|
|
#
?
Apr 23, 2017 04:40
|
|
|
Prediction: i3 will stay the same, just with process improvements, i5 range will be refreshes of the Kaby Lake i7s, and i7 will be 6C/12Ts. Everything i-series gets hyperthreading, and the market segment looks a bit more like it's in direct competition to AMD.
|
|
#
?
Apr 23, 2017 04:44
|
|
|
Intel to introduce "Xeon Gold" and "Xeon Platinum" editions
|
|
#
?
Apr 27, 2017 17:47
|
|
|
Can't wait for the consumer equivalent replacing "Extreme" Editions with "Unsettled" and "Insane" ones. 
|
|
#
?
Apr 27, 2017 18:42
|
|
|
Super Pentium V Turbo
|
|
#
?
Apr 27, 2017 18:58
|
|
|
Xeon Elite Pro Founder's Edition w/Diamond PCIe lane upgrade
|
|
#
?
Apr 27, 2017 19:14
|
|
|
Rastor posted:Super Pentium V Turbo HD Remix
|
|
#
?
Apr 27, 2017 19:20
|
|
|
According to this apparently 2066 socket is identical to 2011 socket. So, can I hope that my Noctua that's been cooling my SB-E (2011 socket) since 2012 will be reusable on the new platform without any modifications? I don't think that i still have my cooler's receipt, since it's been so long.
|
|
#
?
Apr 27, 2017 22:28
|
|
|
The 2011 conversion kit was really cheap too if you didn't remember where you bought it. I think I paid 11 dollars for it. I don't think they're planning to make much money off conversion kit sales even if you do end up needing to buy one.
|
|
#
?
Apr 27, 2017 22:36
|
|
|
craig588 posted:The 2011 conversion kit was really cheap too if you didn't remember where you bought it. I think I paid 11 dollars for it. I don't think they're planning to make much money off conversion kit sales even if you do end up needing to buy one. Oh, I know where i bought it from, i just don't have the receipt. But, i wasn't even aware that they offered the conversion kit for sale. At $11 (hell even at $20), i wouldn't even bother looking for proof that I bought the cooler earlier (though, now looking at Noctua's website, they do say that writing your name/email on a piece of paper and taking a picture of it with the cooler is a good enough proof). But, since they say that 2011 is identical to 2066, i may not even have to do anything, just pop it in.
|
|
#
?
Apr 27, 2017 23:06
|
|
|
Are T-series processors actually binned at all, or just locked down so they can't turbo up to higher TDPs? I'm thinking about doing a ZFS NAS build using a Pentium G4560, but I can also get Pentium G4400Ts for about the same price. It probably won't be running full blast all the time, so if they're binned to idle on lower voltage or something that might be a plus on the T-series. If they're not binned I'd go with the G4560 for race-to-idle. edit: further question: the i3-7100 is only $30 more at Microcenter thanks to the bundle discount. Does the i3 really have anything going for it now that the Pentiums have hyperthreading and AES-NI? Looks like just AVX, and slightly better graphics? Paul MaudDib fucked around with this message at 03:20 on Apr 28, 2017 |
|
#
?
Apr 28, 2017 03:00
|
|
|
Paul MaudDib posted:further question: the i3-7100 is only $30 more at Microcenter thanks to the bundle discount. Does the i3 really have anything going for it now that the Pentiums have hyperthreading and AES-NI? Looks like just AVX, and slightly better graphics? They don't have the baggage of a name that makes them sound like they're from 2005.
|
|
#
?
Apr 30, 2017 23:14
|
|
|
Trip report: my new video encoding server is set up (on the kitchen table - my case hasn't showed up yet). I got a Xeon E5-2650 v3 ES (10C/20T) for $150 off eBay, put it on a GA-X99-UD4 that I got open-box from Microcenter for $60, a Hyper T4 cooler I got for $10 AR from Newegg, and slapped the rest together some leftover components I had laying around. Single-thread performance is pretty pokey compared to the gaming systems I'm used to (only 3 GHz at max turbo/2.3 GHz all-core) but some of that is also probably the terrible 64 GB Transcend SSD I am using at the moment. Works real nice when you hit all the cores though - when doing Handbrake encodes, it's roughly the same performance as my 5820K (4.13 GHz all-core at stock voltages) but only consumes about 2/3 the power (HWMonitor says CPU package power is ~60W vs ~92W for my gaming rig). I was a little worried about temps with that cooler before I knew what the power was going to be like, but I left it encoding overnight and apparently it didn't break 40C package temps and none of the cores broke 45C.
|
|
#
?
May 1, 2017 01:24
|
|
|
T chips do not idle any lower than other desktop chips, though they are remarkably more efficient on load compared to a K or regular desktop chip for the same frequency. I'm guessing they're the chips that can't hit the low idle power targets for laptops but run efficiently when actually doing work which is why they're fairly rare.
|
|
#
?
May 1, 2017 02:59
|
|
|
This should be fun. It appears there is a big security flaw in the Active Management Technology that's part of most of the cpus Intel has sold since 2008. If you don't need AMT make sure it's disabled. If you do need it you'll have to wait for a firmware update.
|
|
#
?
May 2, 2017 02:40
|
|
|
If the reports from others are to be believed, semi accurate is taking a piss poor understanding of a genuine issue and screaming it at the top of their lungs.
|
|
#
?
May 2, 2017 03:34
|
|
|
Drakhoran posted:This should be fun. It appears there is a big security flaw in the Active Management Technology that's part of most of the cpus Intel has sold since 2008. It would be nice if Intel released some specifics on how the attacks works, and specific guidance on if an enterprise PKI setup AMT machine is vulnerable over the network. AFAICT the migitgations involve running Windows software on the local machine.
|
|
#
?
May 2, 2017 04:14
|
|
|
Mr Chips posted:It would be nice if Intel released some specifics on how the attacks works, and specific guidance on if an enterprise PKI setup AMT machine is vulnerable over the network. AFAICT the migitgations involve running Windows software on the local machine. The specifics will be released after patches are made available, since Intel believes going into the specifics right now would make it too easy to exploit systems in the wild.
|
|
#
?
May 2, 2017 04:32
|
|
|
JawnV6 posted:If the reports from others are to be believed, semi accurate is taking a piss poor understanding of a genuine issue and screaming it at the top of their lungs. Well, that is what Charlie does
|
|
#
?
May 2, 2017 04:46
|
|
|
BobHoward posted:Well, that is what Charlie does He wrote some long rant about how Intel's Optane was going to be a flop and the press show they were doing was all smoke and mirrors with phony numbers and tests that journalists wouldn't have the guts to call them on. Then I guess he got invited and pivoted to how because the press briefings he was shown didn't match what was ultimately publicly available that Optane was a huge flop. Now that reviewers got to start poking around with it he's going after the endurance numbers as the reason for it being a flop WhyteRyce fucked around with this message at 05:37 on May 2, 2017 |
|
#
?
May 2, 2017 05:34
|
|
|
fishmech posted:The specifics will be released after patches are made available, since Intel believes going into the specifics right now would make it too easy to exploit systems in the wild. This is the frustrating bit: quote:There are two ways this vulnerability may be accessed please note that Intel® Small Business Technology is not vulnerable to the first issue. The CVE listed 3 months ago now, and Dell/HP still haven't released updated firmware for some affected systems. Intel's vagueness isn't helping us make the most informed decisions.
|
|
#
?
May 2, 2017 06:11
|
|
|
JawnV6 posted:If the reports from others are to be believed, semi accurate is taking a piss poor understanding of a genuine issue and screaming it at the top of their lungs. This is correct. Turns out it's only Q-series (?) boards are affected, and those live in servers. It's still an issue, but slightly smaller than semi accurate wants to make out.
|
|
#
?
May 2, 2017 17:22
|
|
|
|
| # ? Apr 20, 2024 17:06 |
|
|
To mitigate it, you have to unprovision (with ACUConfig or Ctrl+P at boot to get to the menu), then disable the LMS service. If you don't have the password to unconfigure it, you may be able to bypass it through UEFI setup. My Asus board has that option. (It also lets you disable Ctrl+P. Handy.)
|
|
#
?
May 3, 2017 03:23
|
|
