|
AVeryLargeRadish posted:It's not just a cost issue, on smaller dies Intel found that the solder tended to crack after extended heating and cooling cycles, frequently destroying the chip underneath, on larger die chips like the 6+ core i7 and Xeon chips the solder holds up better because of the larger area to distribute heat into. This is me not empty quoting And for more reading http://overclocking.guide/the-truth-about-cpu-soldering/
|
#
?
Feb 9, 2017 17:24
|
|
|
|
| # ? Apr 16, 2024 13:47 |
|
|
While th research on that all seems solid, what would it take for Intel to just make the CPU and GPU die sort of just one bigger die so that the size issue is gone and they can just resolder their lids properly again? Or will this problem pretty much fix itself once 6-8 core consumer chips start becoming the norm?
|
|
#
?
Feb 9, 2017 18:41
|
|
|
What? Chips with iGPUs are already one integrated die.
|
|
#
?
Feb 9, 2017 19:09
|
|
|
EdEddnEddy posted:While th research on that all seems solid, what would it take for Intel to just make the CPU and GPU die sort of just one bigger die so that the size issue is gone and they can just resolder their lids properly again? Die area is money. The bigger the die is, the more that processor cost Intel.
|
|
#
?
Feb 9, 2017 19:16
|
|
|
Oh I thought the iGPU was sort of separate from the GPU in a different way then when I last checked. Nevermind then.
|
|
#
?
Feb 9, 2017 19:17
|
|
|
EdEddnEddy posted:Oh I thought the iGPU was sort of separate from the GPU in a different way then when I last checked. Nevermind then. It was with broadwell  ediT: im wrong, thats the special broadwell cache Don Lapre fucked around with this message at 20:01 on Feb 9, 2017 |
|
#
?
Feb 9, 2017 19:46
|
|
|
Don Lapre posted:It was with broadwell Isn't that just the extra RAM? Crystalwell Double edit: Ah, not Crystalwell, but rather the PCH, it seems the GPU and CPU are still on one die HalloKitty fucked around with this message at 20:06 on Feb 9, 2017 |
|
#
?
Feb 9, 2017 19:49
|
|
|
That pic is a 5775C with the external cache!
|
|
#
?
Feb 9, 2017 19:56
|
|
|
Don Lapre posted:It was with broadwell Right, so I am not crazy as I remember seeing images like that somewhere lol. So if you combined the two in a sense, wouldn't that increase the surface area enough to allow for the soldered lids to work again?
|
|
#
?
Feb 9, 2017 19:58
|
|
|
Gwaihir posted:That pic is a 5775C with the external cache! Yea, i got confused by anandtechs block diagrams.
|
|
#
?
Feb 9, 2017 20:01
|
|
|
Also images such as chips like these for mobile maybe. Either way, i'll be sticking with the larger Xeon rejects for my own systems anyway, but the delid stuff is something to keep tabs on for any customer systems I build that are "gamers" in the future.
|
|
#
?
Feb 9, 2017 20:04
|
|
|
Goddamnit, so it is Crystalwell, although maybe they don't call it that any longer. No idea. The mobile example above has a separate PCH die. More here vv Yeah, I think we just about have it cleared up now, mobile with PCH vs a desktop chip with the extra cache was causing the confusion. Ultimately, GPUs are on the same die as the CPU. HalloKitty fucked around with this message at 20:10 on Feb 9, 2017 |
|
#
?
Feb 9, 2017 20:04
|
|
|
HalloKitty posted:Isn't that just the extra RAM? Crystalwell There were no desktop chips with the PCH built in, that's definitely the external L4 cache. The ultramobile/SOC class chips did have a version that built in the PCH, but they're different looking dies and don't even have heatspreaders at all iirc. http://images.anandtech.com/doci/9320/BDW-H-Map.png e: ^Desktop version die shot Mobile version:  e2: Hm, I'm preeety certain the mobile chips don't have heatspreaders, but now I want to yank an older laptop apart to check.
|
|
#
?
Feb 9, 2017 20:08
|
|
|
I know for a fact that Sandy Bridge M-series and Skylake U-series chips both lack heatspreaders, so it seems like a safe bet that Broadwell/Haswell U/Y models do as well. Haven't actually seen one uncovered in person though.
|
|
#
?
Feb 9, 2017 20:16
|
|
|
EdEddnEddy posted:So if you combined the two in a sense, wouldn't that increase the surface area enough to allow for the soldered lids to work again? Maybe, but (1) Intel seems not at all interested in continuing to pursue further chips with the eDRAM cache because reasons, and (2) it'd only be a short stop-over, as Intel continues to push for 7nm and smaller and smaller dies. On the other hand, I bet AMD and Ryzen is big enough to solder with no problem! 
|
|
#
?
Feb 9, 2017 20:22
|
|
|
DrDork posted:On the other hand, I bet AMD and Ryzen is big enough to solder with no problem! Ryzen's cores are supposedly a touch smaller than Intel's, but since they're going to give us 8 of them for a reasonable price (or so it looks like), then yeah, it will be pretty large overall
|
|
#
?
Feb 9, 2017 20:47
|
|
|
HalloKitty posted:Ryzen's cores are supposedly a touch smaller than Intel's, but since they're going to give us 8 of them for a reasonable price (or so it looks like), then yeah, it will be pretty large overall There's no iGPU on them, which might help.
|
|
#
?
Feb 9, 2017 21:56
|
|
|
For reasons I cannot possibly guess at, Intel has decided to start the hype train on their 8th Generation Core i7-8000 Series: http://wccftech.com/intel-core-i7-8000-series-8th-gen-processors
|
|
#
?
Feb 9, 2017 22:10
|
|
|
Maybe because they realized that the 7000-series is pretty hard to justify in terms of improvements over the 6000-series and is therefore getting basically no hype? I mean, a 15% improvement would be about the jump from 5000->7000, no?
|
|
#
?
Feb 9, 2017 22:29
|
|
|
DrDork posted:a 15% improvement would be about the jump from 5000->7000, no? More like 4000 to 7000 series chips. But yeah a 15% jump would be nice. way late edit: LOL its a 15% jump in sysmark which is a poo poo synth bench nevermind PC LOAD LETTER fucked around with this message at 11:38 on Feb 11, 2017 |
|
#
?
Feb 9, 2017 23:34
|
|
|
PerrineClostermann posted:There's no iGPU on them, which might help. Surely they wouldn't compare including iGPU area
|
|
#
?
Feb 9, 2017 23:44
|
|
|
HalloKitty posted:Surely they wouldn't compare including iGPU area That article can't be including iGPUs- The GT2 iGPU on a Kaby lake chip alone is like 55 mm^2, so that's larger than the 4 actual cores going by that article's table. e: Anand's does a good breakdown of feature sizes in their latest CPU article: http://www.anandtech.com/show/11083/the-intel-core-i3-7350k-60w-review
|
|
#
?
Feb 9, 2017 23:52
|
|
|
HalloKitty posted:Surely they wouldn't compare including iGPU area the iGPU would certainly figure into overall die area, and were were talking about the suitability of various dies for solder vs. other TIM, as I understood it.
|
|
#
?
Feb 10, 2017 00:20
|
|
|
What kind of marketing voodoo is this chart based on? 6th and 7th generation are pretty much the exact same chip with identical IPC, are they assuming 15% higher clocks?
|
|
#
?
Feb 10, 2017 01:02
|
|
|
MaxxBot posted:What kind of marketing voodoo is this chart based on? 6th and 7th generation are pretty much the exact same chip with identical IPC, are they assuming 15% higher clocks? Yes, Kaby Lake chips are clocked a little higher so they're "better".
|
|
#
?
Feb 10, 2017 01:05
|
|
|
MaxxBot posted:What kind of marketing voodoo is this chart based on? 6th and 7th generation are pretty much the exact same chip with identical IPC, are they assuming 15% higher clocks? The old Netburst standby. Moar Hurtz
|
|
#
?
Feb 10, 2017 01:23
|
|
|
Rastor posted:For reasons I cannot possibly guess at, Intel has decided to start the hype train on their 8th Generation Core i7-8000 Series:  I'm happy with my i7 920 -> 7700K jump, but that doesn't make me feel great about buying the drat thing.
|
|
#
?
Feb 10, 2017 01:43
|
|
|
https://twitter.com/tripgabriel/status/829534036951494658 maga
|
|
#
?
Feb 10, 2017 02:13
|
|
|
Eletriarnation posted:I know for a fact that Sandy Bridge M-series and Skylake U-series chips both lack heatspreaders, so it seems like a safe bet that Broadwell/Haswell U/Y models do as well. Haven't actually seen one uncovered in person though.
|
|
#
?
Feb 10, 2017 02:15
|
|
|
Since no one's posted it yet: http://www.guru3d.com/news-story/intel-reacts-to-ryzen-with-new-kaby-lake-core-i5-7640k-and-i7-7740k-processors.html Kind of a dick move by Intel, adding a "high-performance TIM" under the heatspreader. That's marketing for "we actually took our time with these." Still - first news about chips using the new socket...and they're rather underwhelming.
|
|
#
?
Feb 10, 2017 04:04
|
|
|
BIG HEADLINE posted:Since no one's posted it yet: http://www.guru3d.com/news-story/intel-reacts-to-ryzen-with-new-kaby-lake-core-i5-7640k-and-i7-7740k-processors.html Where'd they mention the TIM? A quick skim and a ctrl+f didn't find it, though I am feeling a little tired and might have missed it
|
|
#
?
Feb 10, 2017 05:15
|
|
|
i'm the 800+ pins not being used by gapey lake-x
|
|
#
?
Feb 10, 2017 05:26
|
|
|
PerrineClostermann posted:Where'd they mention the TIM? A quick skim and a ctrl+f didn't find it, though I am feeling a little tired and might have missed it Sorry - it was mentioned on TPU: https://www.techpowerup.com/230474/intel-core-i7-7740k-and-i5-7640k-codenamed-kaby-lake-x-112w-tdp-no-igp "The chips also reportedly feature a high-performance thermal interface material (TIM) under the integrated heatspreaders (IHS)."
|
|
#
?
Feb 10, 2017 05:26
|
|
|
Hey man, compared to mayonnaise, their current TIM is "high performance." They didn't say it was new.
|
|
#
?
Feb 10, 2017 05:31
|
|
|
BIG HEADLINE posted:Since no one's posted it yet: http://www.guru3d.com/news-story/intel-reacts-to-ryzen-with-new-kaby-lake-core-i5-7640k-and-i7-7740k-processors.html Given Intel antics, that HT-enabled i5 will be like $300 (with the iGPU removed from the die to save $$$) for Skylake HEDT mobos that costs $200 and up and they will wondering why the Internet lets out a collective meh.
|
|
#
?
Feb 11, 2017 07:42
|
|
|
Removing the iGPU adds cost, because it changes manufacturing. Most feature differences are just cut e-fuses.
|
|
#
?
Feb 11, 2017 16:19
|
|
|
I think he meant more 'make a kaby lake E/X i5'
|
|
|
#
?
Feb 11, 2017 23:15
|
|
|
If those supposedly leaked Ryzen benchmarks are correct then Intel will need a lot more than "high performance TIM", heh.
|
|
#
?
Feb 11, 2017 23:28
|
|
|
Even if Ryzen rapes Intel like during the Athlon/Pentium 4 days, it is doubtful Intel will lower their prices right? They didn't lower them then as far as I know.
|
|
#
?
Feb 12, 2017 13:34
|
|
|
|
| # ? Apr 16, 2024 13:47 |
|
|
Otakufag posted:Even if Ryzen rapes Intel like during the Athlon/Pentium 4 days, it is doubtful Intel will lower their prices right? They didn't lower them then as far as I know. Have you literally never taken any class related to economics, money, or business, while simultaneously never having any single conversation with anyone about how pricing works?
|
|
#
?
Feb 12, 2017 14:11
|
|
