|
I've never seen temperatures below zero hurt anything
|
#
?
Sep 5, 2017 23:24
|
|
|
|
| # ? May 31, 2024 17:45 |
|
|
Scarecow posted:"A thermal cycle is performed by going from -55 °C to 125 °C while each temperature is hold for 15 minutes." yeah because they didn't want to run the test forever, it's just an intel ad article calm down
|
|
#
?
Sep 5, 2017 23:34
|
|
|
Mofabio posted:yeah because they didn't want to run the test forever, it's just an intel ad article calm down I thought the relevant part was needing to run that extreme thermal cycle up to 300 times before a microcrack (read: not a total cpu failure) occured in the TIM. Sort of proves how resilient it is to begin with. I assume you dabble in physics right, where the rate of reaction doubles for every 10*c increase, halves for decrease. 30 to 70c is far more forgiving than -55c to 125c.
|
|
#
?
Sep 5, 2017 23:45
|
|
|
priznat posted:What happened was after about 4-5 years the device wouldn't boot anymore and then all the drive data was unusable because I couldn't plug it into a pc, had to be in one of the same enclosures. They were EOL'd at that point so it would have been real hard to find a new one. I've run into this multiple times both for myself and clients. The solution is UnRaid, and despite its quirks it's been rock solid and blazing fast, even with nine drives, four dockers and two VMs running on a wheezy old AMD quad core. I will never buy a proprietary NAS again.
|
|
#
?
Sep 5, 2017 23:54
|
|
|
yeah I mean they should have done a sensitivity test but I don't wanna drag anybody looking into CPUs like this instead of producing the hundredth Tomb Raider benchmark. I didn't know those wetting properties of Indium and feel like I learned a lil bit about chips.
|
|
#
?
Sep 5, 2017 23:56
|
|
|
eames posted:T20/T30 are awesome. Mine runs a ton of stuff (NAS, PVR, 8 docker containers, 4 VMs, one of them Windows 10 with a undervolted 1070 for steam remote streaming) at 34W idle. That's less than a 5-bay Synology. A bit offtopic but... I have a T20 running CentOS and started trying to set up a VM with GPU passthrough, but ran into the issue that it disables the iGPU if I add a discrete one. Did you figure out a way to work around that for your system, or are you using two discrete GPUs? I had hoped to save the second full PCIe slot for a SATA controller so I can go over 4 drives, as well as save the power. Other than that, my main complaint about the system is the proprietary power supply - I like the smaller motherboard connector, but it doesn't have enough SATA connections for all the bays in the case which is baffling and I can't replace it with one that does. Eletriarnation fucked around with this message at 00:43 on Sep 6, 2017 |
|
#
?
Sep 6, 2017 00:39
|
|
|
Intel (and AMD) CPUs also last for like 10 years (or really even 20-30 with some systems) so maybe you're not seeing a year of -55 to +125C swings, but maybe you'll see 10 years of 20 to 80C swings. I kind of think the low range is too extreme, but I'm not a materials scientist and there's probably a good reason for getting that low. If it was me I'd do -10C to +125C, but I don't know anything about materials science and thermal cycling testing.
|
|
#
?
Sep 6, 2017 01:42
|
|
|
craig588 posted:there's probably a good reason for getting that low. If it was me I'd do -10C to +125C, but I don't know anything about materials science and thermal cycling testing. Well, for one it helps them sell the narrative that they need to use slop instead of solder on the smaller chips (conveniently also saving a butt-ton of complexity and a bit of cash in the process).
|
|
#
?
Sep 6, 2017 02:13
|
|
|
craig588 posted:Intel (and AMD) CPUs also last for like 10 years (or really even 20-30 with some systems) so maybe you're not seeing a year of -55 to +125C swings, but maybe you'll see 10 years of 20 to 80C swings. I kind of think the low range is too extreme, but I'm not a materials scientist and there's probably a good reason for getting that low. If it was me I'd do -10C to +125C, but I don't know anything about materials science and thermal cycling testing. Materials expand and contract at different rates, so opting for extreme temperature swings is really asking for a failure, because those differences are greater when it's extremely hot or extremely cold. And even then the cpu holds up, only getting warmer due to air gaps in the TIM. We're being critical of the article because it's trying to use fancy words like metallurgy but fails to give an objective conclusion. It's exactly like the analogy of the red lining car. Re: Rate of reaction aka rates of expansion, contraction, heat degradation, IANAS but let's call 20c 1, so 30c would double to 2 etc. The relative difference of normal operating temperatures is far smaller than a drastic swing in either direction let alone both. -60 (0.00390625) -50 (0.0078125 -40 (0.015625) -30 (0.03125 -20 (0.0625 -10 (0.125) 0 (0.25) 10 (0.5) 20 (1) Shutdown/ambient temperature 30 (2) Idle temperature 40 (4) 50 (8) 60 (16) 70 (32) 80 (64) High load 90 (128) Thermal throttle/shutdown temp? 100 (256) 110 (512) 120 (1024) 130 (2048) edit: IOW: it fails 20x harder, not 20x faster. Eyochigan fucked around with this message at 03:12 on Sep 6, 2017 |
|
#
?
Sep 6, 2017 03:03
|
|
|
Intel has always sorta taken the Toyota approach that reliability is all important, and a failure will drive a customer away forever. Of course there are examples where they have hosed that up, but things like the Ryzen memory compatibility issues would be unbelievable shitstorms for Intel. Partly because of their history, and partly because of their market position. Something like sub 10% of K and X series CPU owners ever OC in the first place, so I imagine to intel, the cost and reliability benefits outweigh the extra sales they could potentially make to ultra-nerds craving an extra 200mhz in OC headroom.
|
|
#
?
Sep 6, 2017 03:28
|
|
|
Cygni posted:Intel has always sorta taken the Toyota approach that reliability is all important, and a failure will drive a customer away forever. That was my thinking, but no average consumer drops below ambient temperature, and running at stock should mean lower voltages and temperatures with it. I don't know the exact temperatures but I heard roughly 80-90c when you're overclocked, and dropping to 70c when you de-lid and do the liquid metal hackjob. Solder -should- be slightly better, especially since the process is designed to wet (ie: absorb into surface pores on the metal) into the IHS for full contact instead of paste/gallium sandwich. So I could see 30c to 65c at stock if they simply opted for solder. But they didn't, because.... I would pay a 50 dollar premium for a 7700k that came factory soldered indium et al. edit: I forgot to make my point, I think. If the goal was reliability, why not opt for the cooler more stable process? Because it's "good enough" for stock spec. edit2: there's like no frickin way the solder is going to crack going from 30 to 65 even 70c. point2. Eyochigan fucked around with this message at 03:38 on Sep 6, 2017 |
|
#
?
Sep 6, 2017 03:36
|
|
|
Pretty sure some would pay a $100 premium for a bare die + official shim version.
|
|
#
?
Sep 6, 2017 03:38
|
|
|
Especially if they threw in the "break it via OC and we send you another, once" guarantee that they offered for a while (maybe still do?) with an extra charge.
|
|
#
?
Sep 6, 2017 03:39
|
|
|
BIG HEADLINE posted:Pretty sure some would pay a $100 premium for a bare die + official shim version. I'd pay extra for it. I'd like to direct die mount an aircooler, just because it'd run passive for longer with better contact. I wish Noctua would make a direct die aircooler.
|
|
#
?
Sep 6, 2017 03:41
|
|
|
Eletriarnation posted:Especially if they threw in the "break it via OC and we send you another, once" guarantee that they offered for a while (maybe still do?) with an extra charge. They still offer it - they just don't openly advertise it. https://click.intel.com/tuningplan/ The FAQ for it doesn't specifically mention delidding, even at one point saying that it covers "replacing the processor damaged solely as a result of the overclocking," but then in the next entry they cover their asses by stating: "While we will, under the Plan, replace an eligible processor that fails while running outside of Intel’s specifications, we will not provide any assistance with configuration, data recovery, failure of associated parts, or any other activities or issues associated with the processor or system resulting from overclocking or otherwise running outside of Intel’s published specifications." (emphasis mine) BIG HEADLINE fucked around with this message at 06:48 on Sep 6, 2017 |
|
#
?
Sep 6, 2017 03:41
|
|
|
I wonder if some of Intel's laptop chips are on the same production lines as the desktop chips? That at least would make sense with regards to extreme temperature changes since a laptop is far more likely to see that sort of abuse. Hell, there are laptops that use desktop chips socketed package and all, it's a corner case but there are definitely users who fall into it. I guess Intel could make some special chips just to service that market but the same could be said of overclockers plus overclockers are by definition running the product outside of spec, so I see even less reason to cater to that market than the market of users who subject chips to extreme conditions.
|
|
|
#
?
Sep 6, 2017 06:25
|
|
|
Eletriarnation posted:A bit offtopic but... I have a T20 running CentOS and started trying to set up a VM with GPU passthrough, but ran into the issue that it disables the iGPU if I add a discrete one. Did you figure out a way to work around that for your system, or are you using two discrete GPUs? I had hoped to save the second full PCIe slot for a SATA controller so I can go over 4 drives, as well as save the power. There's an option named "Multi-Monitor Support" (or similar) in the BIOS/UEFI that controls the iGPU when you have an additional graphics card installed. I think it might be disabled by default. Regarding the PSU, it isn't great but it has 78% efficiency at 25W DC which is what my machine idles at. I haven't found a PSU that fits in there and has similarly high low-load efficiency. Even the Corsair SF450 Gold PSU only has 70% at such low loads. You could simply buy a SATA power splitter for your additional drives.
|
|
#
?
Sep 6, 2017 07:36
|
|
|
GRINDCORE MEGGIDO posted:Motherfucking bring it to desktop chips We've been screaming that poo poo ever since Broadwell showed incredible gaming prowess even at its low-ish stock clocks. I have no idea why Intel hasn't obliged.
|
|
#
?
Sep 6, 2017 15:54
|
|
|
eames posted:There's an option named "Multi-Monitor Support" (or similar) in the BIOS/UEFI that controls the iGPU when you have an additional graphics card installed. I think it might be disabled by default. Cool, thanks - I looked in the BIOS but probably glossed over that one due to the nondescript name. I'll have to give it a try. The PSU's not that big of a deal, I just have a 400W Platinum fanless model from my previous try at a home server and wish I could swap that in to see if there's a noticeable difference. Might not be a good idea anyway without a front fan to provide some pressure, though. More concerning in the long term is the idea that if the stock unit ever dies, I can't replace it without some adapter that I may not be able to find or a new OEM unit. AVeryLargeRadish posted:I wonder if some of Intel's laptop chips are on the same production lines as the desktop chips? That at least would make sense with regards to extreme temperature changes since a laptop is far more likely to see that sort of abuse. Hell, there are laptops that use desktop chips socketed package and all, it's a corner case but there are definitely users who fall into it. I guess Intel could make some special chips just to service that market but the same could be said of overclockers plus overclockers are by definition running the product outside of spec, so I see even less reason to cater to that market than the market of users who subject chips to extreme conditions. I thought the common wisdom at this point was that desktop duals were commonly -U or -Y chips that failed to make the power efficiency cut, and desktop quads are the -HQ or E3 Xeon rejects. Do we have a reason to suspect otherwise?
|
|
#
?
Sep 6, 2017 15:58
|
|
|
HalloKitty posted:We've been screaming that poo poo ever since Broadwell showed incredible gaming prowess even at its low-ish stock clocks. I have no idea why Intel hasn't obliged. Because the bigass cache is not as profitable
|
|
#
?
Sep 6, 2017 16:45
|
|
|
BIG HEADLINE posted:They still offer it - they just don't openly advertise it. 1 - "my system wont POST how do I fix it" "what is LLC and is 1.9 a good vcore" 2 - "how do i backup my files or save my bookmarks" 3 - "oh no my psu broke too" 4 - buttmining. or other business disruptions resulting from non working computer. (edit2: probably wrong. but any other type of activity I can think of would fall under issues, and unstable systems/data corruption seems to fit) 5 - "this thing runs too hot" or maybe even "my cpu didn't break enough" 6 - So because this plan is meant to replace the processor damaged solely as a result of the overclocking, and because it includes this disclaimer, I think they'll replace dead cpu's, but they won't help you overclock or take responsibility for any resulting damage. I'd expect worse treatment from intel's legal team honestly. edit: oh right de-lidding was the question. I think the solely part seals it in intel's favour. It could even be considered an outside specification. Eyochigan fucked around with this message at 01:23 on Sep 8, 2017 |
|
#
?
Sep 8, 2017 00:59
|
|
|
So it seems Intel's going to try something different for Coffee Lake: http://wccftech.com/intel-coffee-lake-300-series-z370-z390-chipset-leak/ I guess I'm waiting for the Z390 board.
|
|
#
?
Sep 8, 2017 05:28
|
|
|
BIG HEADLINE posted:So it seems Intel's going to try something different for Coffee Lake: http://wccftech.com/intel-coffee-lake-300-series-z370-z390-chipset-leak/ Really? Jeez, way to kill sales of your locked CPUs right out of the gate if buyers can't get a break on a motherboard.
|
|
#
?
Sep 8, 2017 06:36
|
|
|
BIG HEADLINE posted:So it seems Intel's going to try something different for Coffee Lake: http://wccftech.com/intel-coffee-lake-300-series-z370-z390-chipset-leak/ Dammit! I will never upgrade from my 2500k at this rate.. It's one thing to merely assume there will be something better in 6 months but quite another to have the actual roadmap spelling it out.
|
|
#
?
Sep 8, 2017 06:39
|
|
|
Jesus, if the rumours of Ryzen refresh at 4.3-4.4 GHz are true, AMD's going to wipe the loving floor by 2H 2018.
|
|
#
?
Sep 8, 2017 08:42
|
|
|
They really know how to sour this potential purchase. But I value single thread performance over number of cores, so I guess I stick with them.
|
|
#
?
Sep 8, 2017 09:20
|
|
|
The only thing we know about Z390 at the moment is that it'll offer quality-of-life improvements over Z370, which seems to more or less be a refreshed Z270 with the new socket: Woo, native USB 3.1 Gen 2...just in time for USB 3.2 to become 'a thing.' 
|
|
#
?
Sep 8, 2017 09:50
|
|
|
This rumored Intel Pro series sounds like it could finally bring an option to run Coffee Lake with ECC RAM support. As for the roadmap, it's more or less what I expected. Not allowing Coffee Lake to run on Z170/Z270 is a ballsy move unless there are technical reasons for it.
|
|
#
?
Sep 8, 2017 11:11
|
|
|
Jesus Christ, I've waited so goddamn long to upgrade after Kaby Lake turned out to be a huge nothing burger and Intel hinted at a release this August, and now this, god I hope the Ryzen refresh kicks their rear end.
|
|
#
?
Sep 8, 2017 12:40
|
|
|
Kazinsal posted:Jesus, if the rumours of Ryzen refresh at 4.3-4.4 GHz are true, AMD's going to wipe the loving floor by 2H 2018.
|
|
#
?
Sep 8, 2017 13:28
|
|
|
Anyone venture a guess how big a game improvement I'd get if I replaced my i5-760 with an X3470? I'm mainly playing Overwatch which is extremely bottlenecked by my CPU, and I imagine the hyperthreading alone might help a lot - as well for recording or streaming. I don't know what the OC potential is, but my 760 only managed to clock to ~3.5 GHz (170 BCLK), if I'm reading CPU-Z correctly. I've been waiting eons to replace my CPU, but if Z390 is going to have what those specs indicate, I might as well spend ~€40 for an interim CPU.
|
|
#
?
Sep 8, 2017 17:52
|
|
|
ufarn posted:Anyone venture a guess how big a game improvement I'd get if I replaced my i5-760 with an X3470? Won't increase your FPS at all for Overwatch. Will have less of an FPS drop when playing when recording/streaming.
|
|
#
?
Sep 8, 2017 18:25
|
|
|
We knew the stuff about Z370 and Z390 a while ago. Z370 is essentially a refreshed Z270 (remember that Coffee Lake was not supposed to exist at all before 10nm flopped). Z390 and the rest of the 300 series are the Cannon Lake chipsets, although it looks likely they wont ever be used with Cannon Lake on the desktop. When you look at the stuff Z390 adds... it really looks like a non issue to me?
|
|
#
?
Sep 8, 2017 18:35
|
|
|
I mainly want the new Thunderbolt since the Z390 doesn't looktoo useful except for transfer speed. It's a small thing but I (evidently) don't upgrade too often. 
|
|
#
?
Sep 8, 2017 18:53
|
|
|
ufarn posted:Anyone venture a guess how big a game improvement I'd get if I replaced my i5-760 with an X3470? I remember seeing people getting to a bit under 4GHz with the highest end 1366 Xeon CPUs and I assume the 1156 versions might do similar. 40 dollars for about 10% better performance in CPU limited stuff? Probably not worth it, especially not for games like Overwatch which is very well optimized for lower end systems. craig588 fucked around with this message at 19:16 on Sep 8, 2017 |
|
#
?
Sep 8, 2017 19:13
|
|
|
ufarn posted:I mainly want the new Thunderbolt since the Z390 doesn't looktoo useful except for transfer speed. It's a small thing but I (evidently) don't upgrade too often. You're really going to wait a whole year for a new chipset instead of buying a $50 PCI card?
|
|
#
?
Sep 8, 2017 19:19
|
|
|
I'll see what the release date is, ofc I won't be waiting that long if that's the case. We'll have to see when their desktop CL event is.
|
|
#
?
Sep 8, 2017 19:26
|
|
|
how are they going to integrate wireless? are they going to just stick a shield over part of the package and put some coax ports on there?
|
|
|
#
?
Sep 8, 2017 20:44
|
|
|
Some lukewarm 8700K benchmarks and an Oct 5 release date.
|
|
#
?
Sep 11, 2017 15:10
|
|
|
|
| # ? May 31, 2024 17:45 |
|
|
Dr. Fishopolis posted:You're really going to wait a whole year for a new chipset instead of buying a $50 PCI card? Thunderbolt 3 PCIe cards are available for $50? I hope these end up true since I'd just go ahead and get a R7 1700 and X370 and call it good. I've wanted to see how Coffee Lake's performance would compare and how it might affect prices on Ryzen. Canned Sunshine fucked around with this message at 15:24 on Sep 11, 2017 |
|
#
?
Sep 11, 2017 15:21
|
|
