|
https://www.anandtech.com/show/17221/intel-announces-ohio-fab-complex-2-new-fabs-for-20b-and-space-for-more This is either going to be an insane comeback or an extremely funny image in a few years.
|
# ? Jan 21, 2022 19:38 |
|
|
# ? May 30, 2024 21:32 |
|
movax posted:Most recent Intel street champ was 2600K, IMO. Before that... I want to say the 2.4 or 2.53 Northwood P4s were fairly decent OCs if you had the cooling to keep up (but they also then got their poo poo kicked in my AMD). What about the Q6600? Those were real era-definers that hung on and overclocked great.
|
# ? Jan 21, 2022 19:47 |
|
Cygni posted:https://www.anandtech.com/show/17221/intel-announces-ohio-fab-complex-2-new-fabs-for-20b-and-space-for-more
|
# ? Jan 21, 2022 20:04 |
movax posted:Most recent Intel street champ was 2600K, IMO. Before that... I want to say the 2.4 or 2.53 Northwood P4s were fairly decent OCs if you had the cooling to keep up (but they also then got their poo poo kicked in my AMD).
|
|
# ? Jan 21, 2022 22:36 |
|
movax posted:Most recent Intel street champ was 2600K, IMO. Before that... I want to say the 2.4 or 2.53 Northwood P4s were fairly decent OCs if you had the cooling to keep up (but they also then got their poo poo kicked in my AMD). I remember the later Bloomfield steppings overclocking real well too.
|
# ? Jan 21, 2022 23:03 |
|
Jesus, if Intel's spending $20B for those fabs Micron's $3B they're investing where I am down the road from AWS us-east-1 is comparatively pocket change. Also I wish there was a way to bookmark and remind myself about a post at a scheduled time.
|
# ? Jan 21, 2022 23:15 |
|
Intel has a LOT of money, and if they've de-hosed their processes it makes sense for them to invest heavily. People seem to forget that Intel was dominant not just in CPU design but process technology for a long, long time. They hosed up badly for a long time, but as a company they still have a lot of resources to allocate if they have confidence in something.
|
# ? Jan 21, 2022 23:20 |
|
Not to mention current market conditions and forecasts suggest high end fab capacity is going to be heavily demanded. It's probably not a bad bet.
|
# ? Jan 21, 2022 23:35 |
|
Cygni posted:https://www.anandtech.com/show/17221/intel-announces-ohio-fab-complex-2-new-fabs-for-20b-and-space-for-more I’m originally from Ohio and moved to CA to work at Intel after college. My phone has been blowing up for a week straight now from friends and family. No mom, I’m not moving back home. But really cool to see my employer plop down billions in my home state for a bunch of fabs.
|
# ? Jan 21, 2022 23:36 |
|
VorpalFish posted:Not to mention current market conditions and forecasts suggest high end fab capacity is going to be heavily demanded. Yeah, even if they don't end up retaking the lead in any given segment for the next decade, it seems all but assured that as long as they're somewhat competitive, there will be buyers for every single wafer they can manage to make. In a world where even Samsung has orders out the wazoo for their fabs, it'd be almost fiscally irresponsible not to try to expand capacity. Especially if they're serious about opening up their fabs for custom chip design orders. Taiwan's strategic situation isn't getting any more stable, after all.
|
# ? Jan 22, 2022 02:27 |
|
Cygni posted:Intel sells a 6 P-core overclockable CPU, it’s $100 more. They sell an overclocking chipset, it’s ~$60 more. In Intels own lingo, it’s an “enthusiast” feature with a price premium. As discussed in the vid, they have historically swooped down with a fury on any attempts to overclock frequencies on non-k CPUs or with non-Z chipsets. We will see what they do this time. yeah, i get that and i'm not arguing that they haven't, it's just that -- at the end of the day -- we're talking one specific chip combined with a couple of specific boards. it's loving dumb because intel will end up spending more $ on legal than it'll lose cuz of a nerd's one weird trick.
|
# ? Jan 22, 2022 04:32 |
|
Trolling Thunder posted:intel will end up spending more $ on legal than it'll lose cuz of a nerd's one weird trick. It won't cost them anything, they'll send an angry email or two and it will go away.
|
# ? Jan 22, 2022 05:13 |
|
Yeah. I suppose so. Oh, right—one of those letters made that YouTuber cry and beg for forgiveness because he spoke ill of a chip once. Hilarious stuff.
|
# ? Jan 22, 2022 06:44 |
|
I have a new 12700k and an msi z690 board. I want to OC the cpu since I have a tower radiator on it, but I really don't want to bother with all the ins and outs of OC. Is the OC Genie 4 for the cpu in the msi bios a safe way to get some better performance. I know people will say "Overclocking isn't hard" or "it may not be the absolute best OC you can achieve" but I don't want the best necessarily, just some easy safe gains.
|
# ? Jan 22, 2022 07:14 |
|
Kwolok posted:I have a new 12700k and an msi z690 board. I want to OC the cpu since I have a tower radiator on it, but I really don't want to bother with all the ins and outs of OC. Is the OC Genie 4 for the cpu in the msi bios a safe way to get some better performance. I know people will say "Overclocking isn't hard" or "it may not be the absolute best OC you can achieve" but I don't want the best necessarily, just some easy safe gains. If you’re not gonna get into the nitty gritty you’re not going to even notice a difference. You’ll get 2-300mhz at most even with manually messing with voltage and other settings and it’s not going to improve frame rates or render times in any meaningful or even noticeable way. It’s fun to squeeze the last drop of performance out but you won’t notice an actual improvement beyond marginal score increases on synthetic benchmarks.
|
# ? Jan 22, 2022 13:59 |
|
Kwolok posted:I have a new 12700k and an msi z690 board. I want to OC the cpu since I have a tower radiator on it, but I really don't want to bother with all the ins and outs of OC. Is the OC Genie 4 for the cpu in the msi bios a safe way to get some better performance. I know people will say "Overclocking isn't hard" or "it may not be the absolute best OC you can achieve" but I don't want the best necessarily, just some easy safe gains. Under the stock power limits, your CPU will effectively overclock itself until it hits a power or thermal cap. You can lift the power cap but it's already very high out of the box so the most low-effort way of squeezing out more performance is by undervolting. Unless they've changed it especially for Alder Lake which I strongly doubt, OC Genie will increase the core voltage which is actively counter productive in your case.
|
# ? Jan 22, 2022 14:44 |
|
BlankSystemDaemon posted:Too bad the SMT implementation was a loving disaster to the point that even Intel admitted and removed it, and only reintroduced SMT in Nehalem when they'd finally manged to do it properly. Do you have info on the Netburst SMT implementations? I remember it being a mixed bag under the best of circumstances but would like a refresher with knowledge of better practices this far out.
|
# ? Jan 22, 2022 18:04 |
|
Arzachel posted:Under the stock power limits, your CPU will effectively overclock itself until it hits a power or thermal cap. You can lift the power cap but it's already very high out of the box so the most low-effort way of squeezing out more performance is by undervolting. Unless they've changed it especially for Alder Lake which I strongly doubt, OC Genie will increase the core voltage which is actively counter productive in your case. This was definitely me, for awhile I ran the overclock to get my 9900k to 5GHz all core and besides being able to look at the nice round number on spec screens (except for Task Manager which cheated me out of 30 MHz) I never noticed a single performance benefit, so back to stock I went.
|
# ? Jan 22, 2022 18:56 |
|
Kwolok posted:I have a new 12700k and an msi z690 board. I want to OC the cpu since I have a tower radiator on it, but I really don't want to bother with all the ins and outs of OC. Is the OC Genie 4 for the cpu in the msi bios a safe way to get some better performance. I know people will say "Overclocking isn't hard" or "it may not be the absolute best OC you can achieve" but I don't want the best necessarily, just some easy safe gains. if it takes you 5 minutes to get a stable overclock you will never get those 5 minutes back from any measured performance increase. its just not there.
|
# ? Jan 22, 2022 18:59 |
|
carry on then posted:This was definitely me, for awhile I ran the overclock to get my 9900k to 5GHz all core and besides being able to look at the nice round number on spec screens (except for Task Manager which cheated me out of 30 MHz) I never noticed a single performance benefit, so back to stock I went. I've run my 9900k at 5.0 all core for all of like 3 minutes, I just couldn't be bothered to get the power consumption under control because auto voltage which was some low side of 1.3v shot it to about 225w power consumption and all cores locked to 99C, granted this was running prime95. Even at stock it runs around 190w constantly, but at least the water cooler is able to keep it around 80C in prime. I concluded that there was no way I'd notice a performance uplift in anything from only 6% difference in clock speed and it certainly wasn't going to be worth a 20% increase in power consumption.
|
# ? Jan 22, 2022 19:10 |
Hasturtium posted:Do you have info on the Netburst SMT implementations? I remember it being a mixed bag under the best of circumstances but would like a refresher with knowledge of better practices this far out. It's pretty commonly accepted, I think, that a lot of innovations surrounding CPUs from the mid-90s when that paper came out have been to try and address this memory wall issue, and SMT is no exception. The big problem with the the first SMT implementation that Intel used is that it does significantly worse when it comes to memory benchmarks, as well as other issues outlined in Agner Fogs microarchitectural optimization manual. Interestingly, if you correlate the times Andy Glew worked at Intel with his patents there's a pretty good indication that he was involved in both the first and second implementations (and being responsible for a big portion of every advantage Intel had for a decade or more) - so since they're soon going to stop being patented I'm cautiously optimistic that if he ever does presentations again, he might be able to talk about those times, as I'd love to hear about it. BlankSystemDaemon fucked around with this message at 20:27 on Jan 22, 2022 |
|
# ? Jan 22, 2022 20:24 |
|
BlankSystemDaemon posted:So, one of the things that's always been The Big Problem in terms of performance on CPUs as they've gotten faster has been the memory wall (first, and best, described here). You’ve given me some really cool stuff to read here — thank you for this post!
|
# ? Jan 22, 2022 23:22 |
|
WhyteRyce posted:PC playback has been a minefield since forever and I swear no one actually does it except for the few weirdos left wanting to build HTPCs I definitely fall in to that group, as I want an actual gaming PC in the living room. Not streaming or some garbage, but a fast gaming PC. Then while it's there, may as well use it to play video. Personally I use J. River Media Center. It can use a built in madVR to render, and they're currently working on their own renderer that'll be exclusive
|
# ? Jan 27, 2022 14:20 |
|
DoombatINC posted:Also those few weirdos get very specific models of UHD drives and flash them with hacked firmware to circumvent the DRM and play 4K discs on whatever hardware in VLC anyways What's weird about wanting to do that
|
# ? Jan 28, 2022 13:26 |
|
kliras posted:TIL the Biden administration has yet to pass $52B for semiconductor onshoring. Guess it might still happen, but didn't expect it to take this long. lol expecting this administration to do anything except make life more miserable. Say it with me kids, Better! 👏 Things! 👏 Aren't! 👏 Possible! 👏
|
# ? Feb 5, 2022 08:35 |
|
HWUB reviewed the Alder Lake laptop i7, the i7-12700H, and found that it was the fastest CPU by a good margin within its weight class https://www.youtube.com/watch?v=OYvXx6x3AKc And they found the i3-12100 to be extremely competitive against previous gen six-core parts and very good value as a result https://www.youtube.com/watch?v=2WklVah7ERo
|
# ? Feb 5, 2022 10:05 |
|
gradenko_2000 posted:HWUB reviewed the Alder Lake laptop i7, the i7-12700H, and found that it was the fastest CPU by a good margin within its weight class and just like that the purported benefits of the m1 are completely gone. lmao.
|
# ? Feb 5, 2022 16:27 |
|
carry on then posted:and just like that the purported benefits of the m1 are completely gone. lmao. The main advantage, which is more money for apple, is still there. But yeah it seems beside the singe core efficiency where Intel is pushing the P cores too much, it's a big improvement and up to par with the M1. I'm guessing the slightly lower models would be even better in terms of power and price vs performance. In which case I'll throw my tiger lake yoga out of the window.
|
# ? Feb 5, 2022 20:51 |
|
SwissArmyDruid posted:lol expecting this administration to do anything except make life more miserable. Do you think semiconductor companies are short of money?
|
# ? Feb 5, 2022 21:36 |
|
Didn’t know Biden was refusing to vote for this. What a shame. Especially since he’s head of the legislative branch and all.
|
# ? Feb 5, 2022 21:55 |
|
Trolling Thunder posted:Didn’t know Biden was refusing to vote for this. What a shame. Especially since he’s head of the legislative branch and all. also it passed the house on Friday, though with changes that will need to get worked out with the senate
|
# ? Feb 5, 2022 22:11 |
|
Would be awesome if someone used 1250u UP4 config to make a legit fanless x86 alternative to the MacBook air. Really interested in seeing how performance scales down to 2+8 in a 9w envelope. Probably gonna end up in nothing but tablets or convertibles though...
|
# ? Feb 6, 2022 00:29 |
|
VorpalFish posted:Would be awesome if someone used 1250u UP4 config to make a legit fanless x86 alternative to the MacBook air. But yeah that's what I really want to see too. I have a fanless Core M tablet and while it's not amazing by modern standards, it stays cool when browsing and stuff and gets about half single-thread performance of a brand new 1185G7. With two reasonably fast ores it never really feels slow in normal usage. 2+4 or 8 could be great so waiting for these new chips is the main reason I'm still using this tablet for travel.
|
# ? Feb 6, 2022 01:17 |
|
carry on then posted:and just like that the purported benefits of the m1 are completely gone. lmao. The M1 is typically close, within 10-15%, while using quite a bit less power typically and with 50% of the threads of the 12700H. The 12700H looks like an amazing CPU, but this isn't the huge blowout of the M1 that you think it is. And the M2 is coming later this year so...
|
# ? Feb 6, 2022 18:58 |
|
SourKraut posted:The M1 is typically close, within 10-15%, while using quite a bit less power typically and with 50% of the threads of the 12700H. The 12700H looks like an amazing CPU, but this isn't the huge blowout of the M1 that you think it is. I wouldn't really hold my breath for the M2, everyone who knew what they were doing on the M1 works for Qualcomm now.
|
# ? Feb 6, 2022 19:02 |
|
ConanTheLibrarian posted:Do you think semiconductor companies are short of money? No, but IMO this is a legit problem worth addressing (DoD has been on this for 10+ years now in terms of getting trusted foundries) and we’ve spent money on stupider poo poo!
|
# ? Feb 6, 2022 19:15 |
carry on then posted:I wouldn't really hold my breath for the M2, everyone who knew what they were doing on the M1 works for Qualcomm now. Also, Gerard Williams III, John Bruno, and Manu Gulati started Nuvia (which is now owned by Qualcomm), Jeff Wilcox is at Intel again (he started there originally, if memory serves). BlankSystemDaemon fucked around with this message at 20:07 on Feb 6, 2022 |
|
# ? Feb 6, 2022 19:56 |
|
movax posted:No, but IMO this is a legit problem worth addressing (DoD has been on this for 10+ years now in terms of getting trusted foundries) and we’ve spent money on stupider poo poo! Also it’s pretty common for governments to give incentives to start up factories in places, as long as it is done intelligently (ie not like Foxconn in wisconsin). $52B is a great investment to bring fab capacity back. I have done a little looking but haven’t found anything - did Taiwan subsidize the various tsmc expansions over the years? It would not surprise me and it was a great investment on their part both economically and from a security perspective. I’m sure for the various intel fabs around the world (malaysia etc) there were some significant incentives to move there apart from the cheaper workforce.
|
# ? Feb 6, 2022 20:14 |
|
Hopefully Pat actually builds and uses the fabs instead of treating them as a PR stunt so he can meet presidents like BK did
|
# ? Feb 6, 2022 20:38 |
|
|
# ? May 30, 2024 21:32 |
|
carry on then posted:I wouldn't really hold my breath for the M2, everyone who knew what they were doing on the M1 works for Qualcomm now. M2's design has probably been finalized for at least 12-18 months based on normal CPU architectural timelines. You should really read up more before making absurdist statements. Edit: To provide more background, Apple and TSMC are already evaluating M3 on TSMC's 3nm process, so I don't think anyone should be worried about M2's status. Canned Sunshine fucked around with this message at 22:23 on Feb 6, 2022 |
# ? Feb 6, 2022 22:18 |