|
Bofast posted:If anything, I feel like the 11900K should be the real waste of sand. Less cores than the 10900K and performance should only slightly exceed the 11700K but at a noticable ($140-ish?) price difference. Exactly, the 11700K is only "bad" because it's new but doesn't set a higher standard. It's perfectly fine compared with other comparible CPUs on the market. So yeah, GN should have saved the hyperbole for the 11900K, there's no chance that's going to look good considering the price.
|
#
?
Mar 23, 2021 22:06
|
|
|
|
| # ? Apr 28, 2024 01:23 |
|
|
At the risk of giving SAT flashbacks Turing : Pascal :: Rocket Lake : Skylake Lateral in performance, lateral in MSRP (but actually up in street price), and the new features included (PCIe 4 and AVX-512) are barely used in anything at the moment and fully dependent on future implementation that is likely years away, when you would probably have already replaced the part anyway
|
|
#
?
Mar 23, 2021 22:19
|
|
|
https://www.anandtech.com/show/1657...n-return-of-idf Doubling down on IDM model. Chiplet products incoming.
|
|
#
?
Mar 23, 2021 23:07
|
|
|
Beef posted:Chiplet products incoming.*** ***In 2023
|
|
#
?
Mar 23, 2021 23:31
|
|
|
Gwaihir posted:***In 2023 For desktop, yeah. Looks like Meteor Lake + 7nm in 2023 is the first chiplet part from Intel. For server, Sapphire Rapids is this year and is "tile" based, and I imagine we are gonna see some more chiplet/tile based Foveros mobile parts (despite Lakefield being a big ol flop). Related, intel is reviving tick-tock and aiming for "unquestioned CPU leadership performance"... in 2024. https://www.anandtech.com/show/16574/intel-to-revive-ticktock-model-unquestioned-cpu-leadership-performance-in-20242025
|
|
#
?
Mar 24, 2021 00:18
|
|
|
This is odd I see nothing about drones or game shows or TAM in this news release did Intel hire the right guy?
|
|
#
?
Mar 24, 2021 02:15
|
|
|
WhyteRyce posted:This is odd I see nothing about drones or game shows or TAM in this news release did Intel hire the right guy? I know right, loving distractions
|
|
#
?
Mar 24, 2021 02:31
|
|
|
Wow, Intel is really killing it hard in the PR. I can't remember the last time I saw such a big media push from Intel.
|
|
#
?
Mar 24, 2021 02:48
|
|
|
They need it. Intel is in a pretty bad place on the product side today, so they are probably trying to shift the narrative to the future. Intels desktop launch looks like no improvement, the laptop line is stronger but Tiger Lake H still isn’t out and AMD is taking high margin segments, Ice Lake SP is years behind schedule and only officially launching in April and is launching at Zen 3 Epyc, Cooper Lake isn’t competitive on performance, Optane/Pmem is a mess, DG1 isn’t encouraging, Lakefield flopped, even the network division had some big screw ups with the 2.5G NIC launch. The product side needs a turn around and some good press right about now. E: also forgot that Apple is about to announce their next step in ditching Intel with their own silicon in a few days, lol. Christ Cygni fucked around with this message at 03:10 on Mar 24, 2021 |
|
#
?
Mar 24, 2021 03:07
|
|
|
Doesn't Alder Lake at least look appealing with the new architecture?
|
|
#
?
Mar 24, 2021 03:16
|
|
|
Crosby B. Alfred posted:Doesn't Alder Lake at least look appealing with the new architecture? one would assume that it's going to be better just by dint of the fact that it's 10nm that's been designed from the ground-up as 10nm (as opposed to Rocket Lake's backporting), but I don't really know that the bigLITTLE design is necessarily going to be all that useful / good for the enthusiast space
|
|
#
?
Mar 24, 2021 03:27
|
|
|
Cygni posted:They need it. Intel is in a pretty bad place on the product side today, so they are probably trying to shift the narrative to the future. "We will pursue opportunities with Apple," Gelsinger said on a conference call updating the company's business plans Tuesday evening. Oh word? Is that what you call those youtube adds?
|
|
#
?
Mar 24, 2021 04:25
|
|
|
gradenko_2000 posted:one would assume that it's going to be better just by dint of the fact that it's 10nm that's been designed from the ground-up as 10nm (as opposed to Rocket Lake's backporting), but I don't really know that the bigLITTLE design is necessarily going to be all that useful / good for the enthusiast space Ugh, we're going to have to wait even longer for any decent enhancements? I'm running a i7-8700 which is fine but I really want leap frog into something that'll last me years.
|
|
#
?
Mar 24, 2021 05:21
|
|
|
gradenko_2000 posted:one would assume that it's going to be better just by dint of the fact that it's 10nm that's been designed from the ground-up as 10nm (as opposed to Rocket Lake's backporting), but I don't really know that the bigLITTLE design is necessarily going to be all that useful / good for the enthusiast space I have incredibly little hope for Intel's big.LITTLE efforts and hope I'm wrong. I don't see the needed tight coupling between hardware and software occurring to make it as successful and near flawless as Apple's implementation. Lets look at the top-tier / "halo" PC offerings... Surface, XPS 13 and X1 Carbon (I think, right?). All of them have incentive / OEM relationships to get poo poo right and yet they have silly ACPI bugs, Windows 10 Modern Standby murdering battery life in your backpack and far, far more hardware configurations than an equivalent Apple machine. How the gently caress are they going to manage big.LITTLE without either a fuckton of silicon-level / "hardware" participation (which, lets be real, will never be truly hardware and will instead be software elements that we likely have no visibility into) or actually executing well in conjunction with Microsoft so Win10's scheduler knows what to do?
|
|
#
?
Mar 24, 2021 06:30
|
|
|
like it's possible that the big cores are going to perform well and the target CPU is going to be the 8+0 configurations... but I wouldn't count on itCrosby B. Alfred posted:Ugh, we're going to have to wait even longer for any decent enhancements? I'm running a i7-8700 which is fine but I really want leap frog into something that'll last me years. depending on your budget an i9-10850K might not be a bad idea
|
|
#
?
Mar 24, 2021 06:44
|
|
|
Crosby B. Alfred posted:Ugh, we're going to have to wait even longer for any decent enhancements? I'm running a i7-8700 which is fine but I really want leap frog into something that'll last me years. Probably wait for next Zen with new socket and DDR5. Next year? 8700K isn't that slow unless you want 800fps in CS:GO. At least that is my plan in the midst of this chip shortage.
|
|
#
?
Mar 24, 2021 08:03
|
|
|
Maybe they can win me over in the future, but as of now I have zero desire for a computer CPU with some sort of big.little cores.
|
|
#
?
Mar 24, 2021 14:07
|
|
|
I can see it being cool if the windows scheduler figures it out. I’m sure there will be no problem with that right? Smooth sailing right???
|
|
#
?
Mar 24, 2021 14:16
|
|
|
Sure, say, six months post release maybe! Huge "sit out the first gen" energy on that design. Although really it's so common in other computing spaces that it shouldn't be too radical an adjustment.
|
|
#
?
Mar 24, 2021 14:53
|
|
|
Stupid processor question possibly, but I upgraded about a month ago from an older pc with an i5-6300k to an i9-10850k, buying a custom builder computer so I could get a 3080 along with it for 4k and stuff The computer works great, but one thing I've noticed is that their modest overclocking to 4.7 works super well, but also when the computer is not fully loaded it actually lowers it's clock speed (and presumably consumes less power) which is really cool! Previously I had overclocked my old 6300k to 4.3 GHZ but it would just sit at that the whole time. Admittedly I just used the intel extreme easy overclocker so it probably did a terrible job, but is that a newer feature on processors? It's a really great idea honestly.
|
|
#
?
Mar 24, 2021 16:12
|
|
|
yup, modern processors will "downclock" themselves to very low levels if there's no load for them to work on
|
|
#
?
Mar 24, 2021 16:29
|
|
|
gradenko_2000 posted:yup, modern processors will "downclock" themselves to very low levels if there's no load for them to work on I’m pretty sure his old processor was actually lowering its power state which is separate from the frequency it’s operating at. My 9600 K just sits at 5 GHz all core all the time but core temp can actually show you the power draw of your processor. I am rated for 95 W. At idle im down at 20 W. Cine branch is currently chewing through about 125 W of power. When you start forcing the processor to run at OC frequencies it tends to just stay at those frequencies even though it can also use it’s lower power states at idle. I don’t know what controls whether or not a computer will actually change its clock speed but anytime I give a computer a substantial overclock it doesn’t seem to want to change its clock speeds anymore. Remember that clock speeds and C-sates (power states)are different. If I turn off all of my overclocking my processor will automatically lower its frequency at idle but it doesn’t wanna do it at 5ghz.
|
|
#
?
Mar 24, 2021 17:09
|
|
|
that's how overclocking has worked for quite a while (certainly since at least haswell). I think there might be a setting which forces it to run at the maximum clock all the time but the "normal" way to do it preserves speedstep control, it just lets a higher max multiplier be selected. there's also windows power settings. you can absolutely set a processor power policy that will put the minimum frequency state at 100%. Paul MaudDib fucked around with this message at 17:34 on Mar 24, 2021 |
|
#
?
Mar 24, 2021 17:32
|
|
|
so how'd gelsinger stay clear of the blast radius but keep enough cachet to come back like a hero i dont recall his ouster being... good?
|
|
#
?
Mar 24, 2021 17:56
|
|
|
JawnV6 posted:so how'd gelsinger stay clear of the blast radius but keep enough cachet to come back like a hero I thought he was viewed as a potential next CEO since he was Andy Grove's boy but he was on the outs with Paul because of politics and left because he was intentionally being frozen out and walked into another CEO opportunity. But I guess I also heard he was on the outs because Larabee sucked? Also everyone really hated BK and BS so it's not hard to be seen as good in comparison.
|
|
#
?
Mar 24, 2021 18:02
|
|
|
WhyteRyce posted:I thought he was viewed as a potential next CEO since he was Andy Grove's boy but he was on the outs with Paul because of politics and left because he was intentionally being frozen out and walked into another CEO opportunity. WhyteRyce posted:But I guess I also heard he was on the outs because Larabee sucked?  no it didnt
|
|
#
?
Mar 24, 2021 18:22
|
|
|
Paul MaudDib posted:that's how overclocking has worked for quite a while (certainly since at least haswell). I think there might be a setting which forces it to run at the maximum clock all the time but the "normal" way to do it preserves speedstep control, it just lets a higher max multiplier be selected. Some (older) OCing guides recommended turning off SpeedStep and C-states on the grounds that it could make your overclock slightly more stable, which would lock your CPU to the OC'ed max speed 24/7. But there's no reason to do that these days, as the extra 50Mhz you maybe possibly (probably not) could get out of it isn't worth the hilarious amount of wasted energy use.
|
|
#
?
Mar 24, 2021 18:31
|
|
|
JawnV6 posted:that makes sense. he was head of DHG before skipping out to VMware? DEG before going to VMWare I believe. He was told at VMWare he'd be put on the short list of next CEOs while at the time at Intel I didn't think anyone knew when Paul planned on stepping down quote:
Hey I'm just relaying gossip here
|
|
#
?
Mar 24, 2021 19:22
|
|
|
Elvis_Maximus posted:Stupid processor question possibly, but I upgraded about a month ago from an older pc with an i5-6300k to an i9-10850k, buying a custom builder computer so I could get a 3080 along with it for 4k and stuff This is not a new feature at all and you did your previous OC wrong if it was sitting at max frequency all the time. There were ways to get downclocking to still work with OCs back to at least the Core 2, and the behavior has been very reliable since Sandy.
|
|
#
?
Mar 24, 2021 19:25
|
|
|
K8.0 posted:This is not a new feature at all and you did your previous OC wrong if it was sitting at max frequency all the time. There were ways to get downclocking to still work with OCs back to at least the Core 2, and the behavior has been very reliable since Sandy. There is a difference between sitting at max frequency and sitting at a max power state. All of my overclocks sit at max frequency but they have C states enabled and it absolutely is dialing down the wattage. Speed step is definitely enabled but if you change the core multiplier on my Asus Z390 motherboard the frequency stops fluctuating. Same thing happens on my wife’s Z370 board. Maybe it still fluctuates at a lower OCclock but I’m pushing these things 30% higher than stock so maybe it’s just because I push them hard. I think once you specify a core multiplier it cant automatically change anymore. If somebody wants to tell me how to let my multiplier change on its own while also be pegged at 50x I will definitely listen but it’s something I tooled around with for quite a while in the OC thread and its still 5.0ghz 24/7. It idols under 40°C and it’s only pulling like 20 W of power so I’m not worried about it staying at higher frequencies but I do like understanding how things work. If somebody knows a setting I can touch I would definitely try for the thread. spunkshui fucked around with this message at 21:24 on Mar 24, 2021 |
|
#
?
Mar 24, 2021 21:21
|
|
|
Well it wouldn't be surprising that I did it wrong considering I read a guide for like 3 minutes and then went for it But good to know! It's really neat that it works that way, and seems like a much better idea overall On a related note, that might explain why things started to get all messed up at that overclock after a few months 
Weird Pumpkin fucked around with this message at 22:21 on Mar 24, 2021 |
|
#
?
Mar 24, 2021 22:19
|
|
|
Well the multi changing has been a feature for a long time, and my 6700k does that. Pretty sure speedshift in particular was introduced with skylake? (Before that it was something else that dealt with p-states I think). Here's screenshots of a 10700k at normal 'idle' so webbrowsing etc and at load:  If the frequency stays pegged to max, then a setting as been enabled that prevents downclocking. Perhaps certain motherboards just do that by default if you OC? Shrimp or Shrimps fucked around with this message at 05:27 on Mar 25, 2021 |
|
#
?
Mar 25, 2021 05:22
|
|
|
Yeah if you're pegged at max frequency you flicked a switch somewhere to make it do that, and it would've happily sat at max stock frequency as well. lurksion fucked around with this message at 17:24 on Mar 25, 2021 |
|
#
?
Mar 25, 2021 17:19
|
|
|
Shrimp or Shrimps posted:If the frequency stays pegged to max, then a setting as been enabled that prevents downclocking. Perhaps certain motherboards just do that by default if you OC? The other possibility is that some dumb program is running that's causing the CPU to stay in a high-freq state, even if it doesn't really need to be. I run across that from time to time with Chrome--presumably some website is coded poorly and my CPU ends up sitting at max clocks and like 5% load until I kill it. e; that's part of why I have made a pair of scripts to set the Windows max CPU freq to 50% / 100%, because I can leave it on 50% for daily driving and never notice any performance difference, but it does a good job keeping the power states low and so the cooling system never has much to do. Then I flip it to 100% for the few games I have where it actually matters and I want that full 4.8GHz. DrDork fucked around with this message at 17:25 on Mar 25, 2021 |
|
#
?
Mar 25, 2021 17:22
|
|
|
FuturePastNow posted:Maybe they can win me over in the future, but as of now I have zero desire for a computer CPU with some sort of big.little cores. It turns out it's an awesome thing, actually if done well lol intel's not gonna do it well and even if they do microsoft will probably make it suck somehow
|
|
#
?
Mar 25, 2021 20:39
|
|
|
You know what will happen. A horrible security flaw will be discovered in the little cores and they'll be forced to disable them.
|
|
#
?
Mar 25, 2021 20:41
|
|
|
I want to buy both the 8+0 and 8+8 versions of Alder Lake and hang on to both of them like a jerk until we find out which of them pulls out the stronger responsiveness metrics This will then result in me losing twice in one generation when it's another performance regression and Zen 4 owns
|
|
#
?
Mar 25, 2021 20:47
|
|
|
bull3964 posted:You know what will happen. Other way round would be even more hilarious
|
|
#
?
Mar 25, 2021 20:52
|
|
|
Smol cores are enumerated first. By default game engines pin the main loop to core zero.
|
|
#
?
Mar 25, 2021 21:10
|
|
|
|
| # ? Apr 28, 2024 01:23 |
|
|
HalloKitty posted:Other way round would be even more hilarious Finally, a processor for ants
|
|
#
?
Mar 25, 2021 21:30
|
|