|
I assume the Comet Lake NDA was lifted earlier today. 10900K apparently comes with a stock PL2 of 250W for 56 seconds (9900K was 128W for 28 seconds). I can understand the need for a new socket now, although the overclocking headroom is likely going to be smaller than it was. The new 10 series also allows users to enable/disable Hyperthreading per core as a obligatory Anandtech link: https://www.anandtech.com/show/15758/intels-10th-gen-comet-lake-desktop
|
# ? Apr 30, 2020 16:29 |
|
|
# ? May 30, 2024 10:11 |
|
It's quite dishonest to compare marketing TDP. It's also dishonest to compare raw frequency when there are IPC differences between the two processors. That anand article is junk, and I usually like that site. I might end up buying a new Intel CPU, but I want to see some actual numbers first.
|
# ? Apr 30, 2020 16:32 |
|
They say at the end Intel hasn't shipped any review samples or indicated when or if they intend to do so. I agree it's not up to their usual standards, but they have to put out something on this and they're working with what they've got.
|
# ? Apr 30, 2020 16:39 |
|
Khorne posted:It's quite dishonest to compare marketing TDP. It's also dishonest to compare raw frequency when there are IPC differences between the two processors. That anand article is junk, and I usually like that site. The article specifically talks about the TDP and clocking differences. I thought the article was fine, if not exciting because there isn’t much new to talk about.
|
# ? Apr 30, 2020 17:13 |
|
The computerbase article has more detailed information, but you'll have to machine translate it unless you can read german: https://www.computerbase.de/2020-04/intel-comet-lake-s-cpu-launch-lga-1200/
|
# ? Apr 30, 2020 17:19 |
|
Khorne posted:It's quite dishonest to compare marketing TDP. It's also dishonest to compare raw frequency when there are IPC differences between the two processors. That anand article is junk, and I usually like that site. I mean in practical terms it's pretty predictable what you're going to get here, it's another Kaby Lake with more cores. Sure there is some interesting stuff around the edges, like whether TVB really does anything or whether the revised IHS package marginally improves thermals at all, but assuming those leaked SKUs and clocks were legit you pretty much know what you need to know. The 10700F looks like a legitimately good chip, that moves just-below-stock-9990K performance down to the sub-$300 tier, and Intel still has a small lead in gaming IPC, so if you are willing to put up with the heat and power you have something that will edge out the 3700X in gaming. The 10900K will also still appeal to a certain customer who wants the fastest gaming PC today and doesn't want to wait for Zen3/Rocket, although it would be much more viable at $400 so you don't have to pay a higher price on top of the lower core count vs the 3900X. Other than that, they generally fall into some mix of "not fast enough" or "not cheap enough". Says a lot that they literally didn't even bother to send review samples out. If this had been released in September or October then OK, some people might have been interested but with Zen3 and Rocket coming in literally 6 months, you pretty much should just wait unless you absolutely must build a PC right this minute. Khorne posted:I burst out laughing because they all have hyperthreading now. That particular artificial market segmentation tactic was dirty, and I'm glad some actual competition fixed that practice. AMD's taking that strategy now that they're on top. Their 4000U series basically has 4C4T, 6C6T, 8C8T models as the "basic" R3, R5, R7 models and then if you want SMT that's an additional upcharge. It doesn't quite set off the nerd rage like Intel's Processor DLC since it's not literally a code you can type in, but it's the same net effect. SMT is now an "optional feature" at each tier and you have to pay more to turn it on. Using SMT to segment is a legitimate strategy, AMD couldn't afford to pull their punches with Zen1/Zen+ being comparatively quite weak in many tasks, at least on "enthusiast" parts (2200G and R3 tier parts did exist with this segmentation already, people leave that part off, AMD just didn't do it on the R5 and R7 tier chips). Now that they are on top in the laptop segment they will start looking for new ways to monetize that. Same with the TRX40 socket change, comparatively high TR 3000 prices, etc, where they think they are in the lead they start finding new ways to extract some extra dollars above and beyond "just providing a good product at a good price ". The ideal situation for the consumer is if Intel and AMD are competing heavily in all segments, if AMD just starts dumping on Intel across the board then prices go up and features go away, AMD is not any better about this than Intel. In the Athlon 64 days when AMD was on top, they had $1000+ processors at the top of their stack just like Intel. Paul MaudDib fucked around with this message at 18:18 on Apr 30, 2020 |
# ? Apr 30, 2020 17:39 |
|
eames posted:I assume the Comet Lake NDA was lifted earlier today. That's stock numbers?!?!?! I take it gigantic PSUs are back in style? With 250W-300W of room for GPU, 800-1kW PSUs are looking pretty reasonable.
|
# ? Apr 30, 2020 18:01 |
|
Asrock is now marketing the ability to tell the BIOS to bust the short window TDPs... a thing you could already do on pretty much any Z series board going back to the Z170 days, but didn't have a good ~marketing~ name! https://twitter.com/momomo_us/status/1255850523258785792 TDP tweaking is gonna be huge for the "65w" 10, 8, and 6 cores. Cranking the TDP will make a huge difference in performance if you have aftermarket coolers. If you remember the drama bout 8700 non-Ks in prebuilts having way slower performance and people not understanding why, that was why. Hell, i bet its going to be huge for the 10 cores in general, as my understanding of TVB is that it still is beholden to the short/long term power limits. That is, unless Intel decides to let them wreck the TDP out of the box, which they have mostly only done for K series SKUs the last few gens. To be fair, AMD is playing the same game since regulators don't seem to care about the blatant TDP fuckery, but its still a stupid game.
|
# ? Apr 30, 2020 19:12 |
|
The AT Z490 motherboard round up: https://www.anandtech.com/show/15723/the-intel-z490-motherboard-overview The bottom of the price scale is the same as the Z390 launch, about $150. The top of the price scale though is now astronomical, and for Asus at least, a lot of the established popular SKUs are gaining like $100 on the prices. The Maximus Hero is $399 vs $299 last generation. It is actually pretty different when you start comparing features, but that same big spread in price/features we saw in the x570 launch has def carried over to Z490. The highest end boards are going to be over $1,300 lol. Cmon.
|
# ? Apr 30, 2020 20:39 |
Cygni posted:The AT Z490 motherboard round up: Having said that, the Supermicro based board with PLX to provide 4x PCI-ex x8 and 1x PCI-ex x1 has me insanely intrigued for my workstation, even if it's 2 cores less than AMD 3900X.
|
|
# ? Apr 30, 2020 21:01 |
|
https://www.anandtech.com/show/15758/intels-10th-gen-comet-lake-desktopquote:Users wanting the 10-core 5.3 GHz will need to purchase the new top Core i9-10900K processor, which has a unit price of $488, and keep it under 70 ºC to enable Intel’s new Thermal Velocity Boost. Not only that, despite the 125 W TDP listed on the box, Intel states that the turbo power recommendation is 250 W – the motherboard manufacturers we’ve spoken to have prepared for 320-350 W from their own testing, in order to maintain that top turbo for as long as possible.
|
# ? Apr 30, 2020 23:16 |
|
D. Ebdrup posted:The lack of ECC is so loving frustrating! Come on, Intel. What of intel's behavior in the last 25 years led to you to have even an inkling that they will ever support ECC on a desktop chip until some Conroe-level platform shift happens again, and even then, market segmentation.
|
# ? Apr 30, 2020 23:42 |
|
Amd making it a required checkbox or ddrN making it required to function are the only reasons I can think of
|
# ? Apr 30, 2020 23:46 |
|
taqueso posted:Amd making it a required checkbox or ddrN making it required to function are the only reasons I can think of For what its worth, ECC and ECS are going to be on-die with DDR5. Of course that doesn't force anybody to use it within their memory controller, but the old "its an added expense on the PCB and added work for validation" excuses won't be valid anymore.
|
# ? May 1, 2020 00:21 |
|
Edit: sorry, wrong thread.
Kane fucked around with this message at 01:16 on May 1, 2020 |
# ? May 1, 2020 01:08 |
|
Cygni posted:For what its worth, ECC and ECS are going to be on-die with DDR5. Of course that doesn't force anybody to use it within their memory controller, but the old "its an added expense on the PCB and added work for validation" excuses won't be valid anymore. The bigger expense is the requirement you spend 12.5% more on memory for the ecc die. Since oems drive development why take on an extra cost they don't want (plus power usage etc) I agree that with ecc on die there's less of a need but it's still gonna be more traces and of course market segmentation
|
# ? May 1, 2020 08:17 |
|
Computerbase figured out that all -K SKUs and the 10400/-F seem to use the 10C Die with a soldered IHS, while everything below that seems to use the 6C Die. They updated their article. They also note that the 10700K at 95W TDP-down has a 100 MHz lower baseclock than the 9900K, which would make it a less efficient CPU. That would be strange.
|
# ? May 1, 2020 09:34 |
|
Maybe I just missed it, but did Intel give a release date for the new chips?
|
# ? May 1, 2020 09:47 |
I mean, it's not as if all of the other storage tiers available in the form of caches don't use ECC. Harddisks use it too. System memory is the only exception, all because some PC clones wanted to compete on cutting costs the most.Crunchy Black posted:What of intel's behavior in the last 25 years led to you to have even an inkling that they will ever support ECC on a desktop chip until some Conroe-level platform shift happens again, and even then, market segmentation. Cygni posted:For what its worth, ECC and ECS are going to be on-die with DDR5. Of course that doesn't force anybody to use it within their memory controller, but the old "its an added expense on the PCB and added work for validation" excuses won't be valid anymore.
|
|
# ? May 1, 2020 10:07 |
|
Without looking up specs, is it possible to rank intel CPUs across generations by processing power based solely on their model numbers? Assuming cooling is not a factor. And including desktop and laptop parts.
|
# ? May 1, 2020 23:10 |
|
Only if you're got what all the suffixes mean memorized and happen to know what the core count for given model is. Generally, for a given I#, higher numbers (the 4 that come after), higher equals better. But you have to sort out the mobile parts since unless they ended in Q they were dualcores for the longest time, regardless of what I# they were. Also changes in thread count between 8th and 9th gen muddy it a bit, depending on workload. And even then I'm not sure how lower end versions of a new gen compare to higher ones from the previous. Probably poorly. All this is to say, no, except in the most technical sense. Fantastic Foreskin fucked around with this message at 23:22 on May 1, 2020 |
# ? May 1, 2020 23:20 |
|
You’d think putting the core count into the model number would be useful.
|
# ? May 1, 2020 23:36 |
|
For gen 1-7, non mobile parts i3 was 2C/4T, i5 was 4C/4T, and i7 was 4C/8T. For 8th gen i3 was 4C/4T, I5 6C/6T, I7 6C/12T. For 9th gen its I3 4C/4T, I5 6C/6T, I7 8C/8T, and I9 8C/16T. Mobile is a whole 'nother fettle of kish. For 7th gen and before I think (but haven't verified) anything without a Q is 2C, and with is 4C, no idea about SMT on the models across the gens. 8th and 9th gens added cores and I don't know what has what anymore.
|
# ? May 1, 2020 23:46 |
|
Shaocaholica posted:You’d think putting the core count into the model number would be useful. Useful for users... but the purpose of the model naming isnt to help users, its to sell CPUs.
|
# ? May 2, 2020 00:52 |
|
Some Goon posted:Only if you're got what all the suffixes mean memorized and happen to know what the core count for given model is. Generally, for a given I#, higher numbers (the 4 that come after), higher equals better. But you have to sort out the mobile parts since unless they ended in Q they were dualcores for the longest time, regardless of what I# they were. Also changes in thread count between 8th and 9th gen muddy it a bit, depending on workload. And even then I'm not sure how lower end versions of a new gen compare to higher ones from the previous. Probably poorly. No, Q has always meant quad-core. Shaocaholica posted:Without looking up specs, is it possible to rank intel CPUs across generations by processing power based solely on their model numbers? Assuming cooling is not a factor. And including desktop and laptop parts. Short answer: no. Long answer: processor performance heavily depends on the power budget it's allowed, even in the desktop market. A 6700T is going to get pantsed by a 4790K. Generally in the desktop market if you rank by power budget first, and then you sort by i3/i5/i9, then by ID number (meaning, generation and model) you will get something vaguely accurate but not 100%. So like, look at the end of the name first, then the first part, then the middle bit, cause that makes sense. Laptops, all bets are kinda off. Generally high power will beat low power but they also moved to higher core counts at some points in the various series and that makes a difference too.
edit: now for the really fun one: try to find some coherent way to rank Xeon processors by performance. Even within E5 and within the new bronze/silver/gold/platinum rating schemes it makes no goddamn sense. Good luck without just looking at a table. Paul MaudDib fucked around with this message at 02:54 on May 2, 2020 |
# ? May 2, 2020 02:49 |
|
ConanTheLibrarian posted:Maybe I just missed it, but did Intel give a release date for the new chips? Pre-order May 5th, review embargo and availability May 20th. *huge jerkoff motion* Said it before when AMD did this with the 3000 series, but letting pre-orders start weeks before your review embargo ends is scammy bullshit, which is right in line with how both of these dogshit companies operate.
|
# ? May 2, 2020 02:57 |
|
Paul MaudDib posted:No, Q has always meant quad-core. "Unless", my man.
|
# ? May 2, 2020 04:24 |
|
Cygni posted:Pre-order May 5th, review embargo and availability May 20th. *huge jerkoff motion* anyone preordering this turd pushed out by the 14nm fabs at intel deserves what they get
|
# ? May 2, 2020 05:54 |
|
Some Goon posted:"Unless", my man. Are there actually "unless" on that though? I actually don't remember any non-quad Q chips.
|
# ? May 2, 2020 07:58 |
|
the dual-cores were all called Core 2 Duo Exxx the quad-cores were all called Core 2 Quad Qxxx there was one SKU that was a dual-core, but was sold under the Core 2 Extreme branding, which was the X6800 all the other Core 2 Extreme chips were designated QXxxx and were quad-cores (there was also four Core 2 Solo SKUs, the U2100, U2200, SU3300, and SU3500, which were single-core CPUs)
|
# ? May 2, 2020 08:18 |
|
In case one tiny-rear end fan on X570 mainboards didn't do it for you, you can get Z490 mainboards with two: https://post.smzdm.com/p/apz36g99/ orcane fucked around with this message at 11:50 on May 2, 2020 |
# ? May 2, 2020 11:46 |
|
Paul MaudDib posted:Are there actually "unless" on that though? ? It says they were dual cores unless it had a Q. It could use some more commas I'll give you.
|
# ? May 2, 2020 14:43 |
|
orcane posted:In case one tiny-rear end fan on X570 mainboards didn't do it for you, you can get Z490 mainboards with two: "phantom gaming" and no PCIe slot!? WTF, ASRock?
|
# ? May 2, 2020 15:59 |
|
Isn't that a PCIe slot hidden behind the m.2 riser card?
|
# ? May 2, 2020 16:02 |
|
Yeah it's a "regular" mainboard otherwise (with TB3 and other stuff stacked on top): I just loved that they stuck two tiny fans on there. The MSI and Asus m-ITX boards apparently also use a fan for the VRM stack. Gotta run those clocks somehow, after all.
|
# ? May 2, 2020 16:07 |
|
gradenko_2000 posted:the dual-cores were all called Core 2 Duo Exxx I didn’t specifically say but the game I was imposing should have been limited to the ‘i’ model numbers after core2.
|
# ? May 2, 2020 16:10 |
|
.
sincx fucked around with this message at 05:55 on Mar 23, 2021 |
# ? May 2, 2020 16:28 |
|
orcane posted:Yeah it's a "regular" mainboard otherwise (with TB3 and other stuff stacked on top): Check out my stealth black build *turns on RGB*
|
# ? May 2, 2020 16:35 |
|
Ah, okay. That doesn't make it *not* an idiotic mainboard.
|
# ? May 2, 2020 16:50 |
|
|
# ? May 30, 2024 10:11 |
|
orcane posted:In case one tiny-rear end fan on X570 mainboards didn't do it for you, you can get Z490 mainboards with two: luv my thicc boi motherboards! Neat they put full size dimms on there too
|
# ? May 2, 2020 20:27 |