|
What's the context of this
|
# ? May 11, 2018 18:17 |
|
|
# ? May 9, 2024 22:59 |
|
VostokProgram posted:What's the context of this
|
# ? May 11, 2018 21:59 |
|
VostokProgram posted:What's the context of dis
|
# ? May 13, 2018 05:46 |
|
Argh, what's that 8 core cpu later this year gonna be??? Are they going to go x+2 cores on everything again, or will they find a stupid way to make this an inconvenient choice somehow? I can't believe intel so easily adds cores when they haven't all these years. I wish we knew already.
|
# ? May 13, 2018 10:49 |
|
I'd rather want to know if there's going to be a Cascade Lake-X this year, or if those rumors about the soldered-IHS Skylake-X refresh are true.
|
# ? May 13, 2018 11:40 |
|
Intel finally announced Z390. The leaks were correct, it's just Z370 plus some extra I/O features (mainly 802.11ac, Bluetooth 5 and USB 3.1gen2).
|
# ? May 13, 2018 15:29 |
|
lllllllllllllllllll posted:Argh, what's that 8 core cpu later this year gonna be??? Are they going to go x+2 cores on everything again, or will they find a stupid way to make this an inconvenient choice somehow? I can't believe intel so easily adds cores when they haven't all these years. I wish we knew already. I expect it to perform just like a 8700K with two additional cores and slightly hotter. TDP will go up but the larger die size, better TIM*, process improvements and binning will counter that to some degree. The ringbus architecture seems to work ok with up to 10 cores (see i7-6950X). *Der8auer mentioned in one of his videos that Intel is experimenting with liquid metal TIMs
|
# ? May 13, 2018 15:35 |
|
The eight-core 14nm+++ CPUs are probably going to cost about as much as the 7820X. So long as they're keeping with 14nm, the addition of two more cores halves the amount of viable cores they get from each wafer. And I'd fathom most people jumped on the six-cores early, with a thought to go octa with Ice Lake or later. Buying into Z390 is effectively buying an end-of-life product. Sky-X is at least due for a *single* refresh, even if it probably won't be anything stellar or ground-breaking. And if the CPUs *don't* cost as much as the 7820X, the motherboard makers will effectively *make* the combo cost as much as Sky-X because I'd imagine the Z390 isn't going to be a stellar seller, and thus it won't be made in as high of a volume as the Z370 was. BIG HEADLINE fucked around with this message at 16:52 on May 13, 2018 |
# ? May 13, 2018 16:50 |
|
BIG HEADLINE posted:The eight-core 14nm+++ CPUs are probably going to cost about as much as the 7820X. I talked to the OCN Gigabyte rep, and he stated Gigabyte and presumably other manufacturers held off on high end Z370 to focus their effort on Z390. It's why there is no Z370 Maximus Extreme or Gaming 9 or SuperCarrier, and I expect to see these with Z390. Hopefully that points towards a bit more staying power than you might guess. I've personally been waiting for Z390 just because there are no motherboards that really tick all the boxes for me with Z370, and I am happy to pay more for a top end PLX flagship board.
|
# ? May 13, 2018 17:00 |
|
How necessary it is to upgrade the hardware of my htpc? It's 3,6ghz i7 920 with 12gb ddr3. Everything works fine and I really don't want to spend money unless absolutely necessary. If necessary, what would be at least as good, but with as low a price tag as possible? I'd apparently need a new cpu, mobo & ram.
|
# ? May 13, 2018 17:03 |
|
Ihmemies posted:Everything works fine and I really don't want to spend money unless absolutely necessary. That's way more powerful than my HTPC and I don't feel limited at all. The only major thing you'd have to gain is power use, but it'll never make up the extra cost of spending more on new hardware.
|
# ? May 13, 2018 17:17 |
|
craig588 posted:That's way more powerful than my HTPC and I don't feel limited at all. The only major thing you'd have to gain is power use, but it'll never make up the extra cost of spending more on new hardware. I know. It's just about spectre&meltdown and that the platform won't get patches from Intel. And people just discussed earlier in this thread that running windows on vulnerable hardware is a bad idea.
|
# ? May 13, 2018 17:37 |
|
Even though this is the Intel thread, an AMD 2400G setup would probably suit you just fine.
|
# ? May 14, 2018 07:35 |
|
Ihmemies posted:I know. It's just about spectre&meltdown and that the platform won't get patches from Intel. And people just discussed earlier in this thread that running windows on vulnerable hardware is a bad idea. Microcode is only required for Spectre, meltdown is mitigated with kernel changes though its performance impact is mitigated somewhat with newer CPU gens. Neither of the attack vectors are terribly useful for an attack on an HTPC system, there's not much in the way of sensitive memory to scrap out of one.
|
# ? May 14, 2018 15:17 |
|
I do wonder why to help get a partial handle on 86/8700K thermal issues, why they haven't released an xx50K variant that has the iGPU disabled. If the 8086K has that, I might bite early - but I keep hearing rumors that there might be 'surprising' news at Computex about Z390 and the 14nm+++ octacores.
|
# ? May 15, 2018 02:00 |
|
BIG HEADLINE posted:I do wonder why to help get a partial handle on 86/8700K thermal issues, why they haven't released an xx50K variant that has the iGPU disabled. If the 8086K has that, I might bite early - but I keep hearing rumors that there might be 'surprising' news at Computex about Z390 and the 14nm+++ octacores. Is that really necessary? I can't imagine the iGPU draws very much power when not in use.
|
# ? May 15, 2018 02:38 |
|
VostokProgram posted:Is that really necessary? I can't imagine the iGPU draws very much power when not in use. Very likely not, but why have something on the chip that 99% of people aren't going to use? It'll be an Anniversary Edition for enthusiasts. The 2550K had a 100Mhz higher clock speed than the 2500K had, so obviously it frees up *something* within a given thermal budget.
|
# ? May 15, 2018 07:21 |
|
Dream scenario: AMD drops a big increase in clockrates on 7nm next year (say, 4.7 GHz all-core turbo) and Intel is forced to respond with their own increases to all-core turbo and soldered lids. I suppose that in this scenario TDPs are also going up as well, but apart from grumbling from the tech media nobody really cares as long as the performance is there. And really, that's just kind of the reality of 8-core processors in a world where Moore's Law is dead. Either clockrates are going down, or TDP is going up. I suppose while I'm wishing I'd like full-speed AVX2 support on Ryzen as well. Clockrates and AVX are really the only grumbles left with Ryzen at this point. Paul MaudDib fucked around with this message at 09:08 on May 15, 2018 |
# ? May 15, 2018 09:03 |
|
I hope we'll see a return of the L4 eDRAM cache or some variation thereof. That'd improve performance across the board, even singlethread, without major downsides besides cost. All the eDRAM equipped CPUs seem to age exceptionally well and Intel could realistically release a (expensive) soldered 8-core 4.8 GHz CFL refresh with 256MB L4 cache on 14nm. The only problem is that they wouldn't be able to match that on 10nm for years to come.
|
# ? May 15, 2018 09:33 |
|
Paul MaudDib posted:Dream scenario: AMD drops a big increase in clockrates on 7nm next year (say, 4.7 GHz all-core turbo)
|
# ? May 15, 2018 13:11 |
|
Paul MaudDib posted:Dream scenario: AMD drops a big increase in clockrates on 7nm next year (say, 4.7 GHz all-core turbo) and Intel is forced to respond with their own increases to all-core turbo and soldered lids. Would you be happy with 2x7700K performance at 2x7700K TDP? Somewhere around 140W, TDP does start to be a real problem. There's a reason server cooling is so obnoxiously loud.
|
# ? May 15, 2018 14:54 |
|
Hoping they don't solder them, because: - internet rage - direct die cooling.
|
# ? May 15, 2018 15:10 |
|
GRINDCORE MEGGIDO posted:- direct die cooling. Too many people used to crack their CPUs; they're never going to do this.
|
# ? May 15, 2018 15:46 |
|
HBM + cpu dies on mcm >>> more cores
|
# ? May 15, 2018 16:13 |
|
eames posted:I hope we'll see a return of the L4 eDRAM cache or some variation thereof. That'd improve performance across the board, even singlethread, without major downsides besides cost. They aren't in a position to increase die size unless they move to the glue model they laughed at. They have 10nm 2core skus with no igpu, which should be telling about how 10nm is going.
|
# ? May 15, 2018 16:21 |
|
eames posted:I hope we'll see a return of the L4 eDRAM cache or some variation thereof. That'd improve performance across the board, even singlethread, without major downsides besides cost. intel honestly just needs to look at 5nm and beyond.
|
# ? May 15, 2018 16:27 |
|
Inept posted:Too many people used to crack their CPUs; they're never going to do this. I don't see them doing that either. Shame. I'd personally take a non soldered IHS on desktop chips, even 8c, so it can be popped off and direct die cooled. I'll change that attitude instantly when I break one
|
# ? May 15, 2018 17:22 |
|
Lmao https://twitter.com/TMFChipFool/status/996477548757385216?s=19 https://twitter.com/TMFChipFool/status/996477810481991680?s=19
|
# ? May 15, 2018 20:55 |
|
I don't know what you're worried about, 10nm Cannon Lake is already shipping... ... as a sexy dual core, non-iGPU Core i3 in a $399 Lenovo equipped with a 15" 1.366 x 768 screen, 4GB RAM and a 5400rpm boot drive.
|
# ? May 15, 2018 21:29 |
|
eames posted:I don't know what you're worried about, 10nm Cannon Lake is already shipping...
|
# ? May 15, 2018 21:36 |
|
mystes posted:Why are they even bothering to ship these processors? Recoup any of the cost that they can, probably.
|
# ? May 15, 2018 21:40 |
|
mystes posted:Why are they even bothering to ship these processors? Two plausible explanations : They promised shareholders to ship Cannon Lake in late 2017 and technically that's what they're doing with this. Maybe some executives and engineers are getting a bonus out of this. Recycling these CPUs to improve 10nm yields would be more expensive than
|
# ? May 15, 2018 21:44 |
|
MaxxBot posted:Lmao dear lord
|
# ? May 15, 2018 21:47 |
|
10W 14nm Atom beats that 15W 10nm Cannon Lake by ~10% in Cinebench Multicore. https://twitter.com/Dayman58/status/995290335910727681 (yes it's quad vs dual but still)
|
# ? May 15, 2018 21:57 |
|
Gyrotica posted:Recoup any of the cost that they can, probably.
|
# ? May 15, 2018 22:01 |
|
eames posted:10W 14nm Atom beats that 15W 10nm Cannon Lake by ~10% in Cinebench Multicore. Atom is the future, those new Denverton Atoms are looking a hell of a lot more appealing than the Skylake Xeon-Ds.
|
# ? May 15, 2018 22:04 |
|
eames posted:10W 14nm Atom beats that 15W 10nm Cannon Lake by ~10% in Cinebench Multicore. e: nvm, thought you were implying that the 6006U is an Atom but I misunderstood. I'm not super surprised considering that at this point Atom's IPC is comparable to Core 2. It makes perfect sense for Intel to sell these as cheap dualcores first if yields are sucking that much, they can at least recoup some money instead of throwing out all that expensive silicon. Eletriarnation fucked around with this message at 22:58 on May 15, 2018 |
# ? May 15, 2018 22:54 |
|
So what I'm reading from all this is I should sacrifice a goat to make sure my 2500k@4.2 lasts another 18-24 months?
|
# ? May 15, 2018 22:58 |
|
Oh, the processors are what were supposed to be normal i-whatever 4+ core 10nm processors with working gpus? I misunderstood and thought they were something crappier than that, but in that case it makes sense that intel has them lying around and wants to recoup the costs by selling them. That must be really terrible yield though.
|
# ? May 15, 2018 23:01 |
|
|
# ? May 9, 2024 22:59 |
|
mcbexx posted:So what I'm reading from all this is I should sacrifice a goat to make sure my 2500k@4.2 lasts another 18-24 months? Buy a Threadripper. Threaderipper Gen2 will be a powerhouse, and Threadripper gen1 is down to ~$700 for 16C/32T.
|
# ? May 15, 2018 23:02 |