|
WhyteRyce posted:I’m 100% certain that 90% of the people in my semiconductor physics classes didn’t actually know what the stuff was for. Like they tried to dive right in to saturation and calculations but at a basic level didn’t know what a transistor was used for even after passing the class To be fair there is an entire segment of the EE populace that can get away with thinking transistors == solid-state switches for a long time. Then their poo poo breaks and they discover everything is analog in the end.
|
#
?
Aug 25, 2022 23:33
|
|
|
|
| # ? Jun 6, 2024 08:57 |
|
|
Passburger posted:This might sound silly, but make sure you don't have any powerbricks close to the DAC, usb cable or the headphone cable. I had noise for the longest time and thought it was motherboard related, until I started moving things around, turns out it was a powerbrick from my monitor being too close to my xlr cable. Anything XLR should be balanced, so if that picks up audible interference from a power brick that is some pretty crazy noise. Or maybe just cheap gear that uses a XLR connector without actually doing balanced audio, to appear better than it is? Also USB cables are not really susceptible to incidental noise -- it's a digital connection, it pretty much either works or doesn't.
|
|
#
?
Aug 25, 2022 23:35
|
|
|
Klyith posted:Also USB cables are not really susceptible to incidental noise -- it's a digital connection, it pretty much either works or doesn't. They can introduce noise, however. I can reliability get a noisy audio signal out of my phone if I'm using wired headphones and plug the phone into my computer to charge while listening. Although that's more the power coming from the computer than the cable itself. Still.
|
|
#
?
Aug 26, 2022 00:16
|
|
|
I don't know how much we should trust this as I've become quite skeptical of the source (greymon55), but the results would be surprising if true: https://videocardz.com/newz/amd-ryzen-7700x-7600x-rumored-cinebench-r23-scores-are-in It suggests a much larger single-threaded cinebench score than expected and a much lower multi-threaded score than expected. I thought the idea was that these CPUs would not have much IPC gains but they'd be able to achieve much higher all-core boost clocks.
|
|
#
?
Aug 26, 2022 05:37
|
|
|
Dr. Video Games 0031 posted:I don't know how much we should trust this as I've become quite skeptical of the source (greymon55), but the results would be surprising if true: https://videocardz.com/newz/amd-ryzen-7700x-7600x-rumored-cinebench-r23-scores-are-in I guess it's feasible that you would see that sort of result if they run hotter than expected, enough that the test unit wasn't able to sustain the max boost clocks on all core loads without throttling, but managed well enough on single core?
|
|
#
?
Aug 26, 2022 08:25
|
|
|
Dr. Video Games 0031 posted:I don't know how much we should trust this as I've become quite skeptical of the source (greymon55), but the results would be surprising if true: https://videocardz.com/newz/amd-ryzen-7700x-7600x-rumored-cinebench-r23-scores-are-in Comparing the purported 7700x results to the previous generation 8c/16t parts does show a larger percentage improvement for mt than st? Also not clear what the clockspeed v ipc thing has to do with mt v st. Higher all-core clocks doesn’t preclude even higher clocks for fewer cores.
|
|
#
?
Aug 26, 2022 13:38
|
|
|
this allusion meant posted:? Yes, but the ST improvement over Zen 3 is still greater than expected and the MT improvement is less than expected. I figured everyone was expecting a larger gap between the two, with Zen 4 struggling against Intel at ST but being dominant at MT.
|
|
#
?
Aug 26, 2022 14:18
|
|
|
i've been idly reading up on thin clients and noticed this is a pretty neat piece of kit https://www.amazon.com/Beelink-GTR5-5900hx-Windows-Computer/dp/B09S3SX5LT 5900HX + 32GB RAM + 500GB NVME + Wifi 6E + 2x2.5GbE
|
|
#
?
Aug 26, 2022 14:23
|
|
|
Dr. Video Games 0031 posted:Yes, but the ST improvement over Zen 3 is still greater than expected and the MT improvement is less than expected. I figured everyone was expecting a larger gap between the two, with Zen 4 struggling against Intel at ST but being dominant at MT. The MT performance in that leak lines up with the ~40-50% improvement AMD showed off in the announcement, doesn't it?
|
|
#
?
Aug 26, 2022 16:22
|
|
|
shrike82 posted:i've been idly reading up on thin clients and noticed this is a pretty neat piece of kit Hell of a thin client - all it needs is a more serious GPU and they could sell it as a workstation.
|
|
#
?
Aug 26, 2022 16:47
|
|
|
A week or so ago, mdxi very kindly gifted me a used 3600, which I used to replace my 2600X. As a result, I have a free 2600X to give away. It will come with an unused stock cooler in a box for a 5600G, and I will ship it to you anywhere in the lower 48. First PM requesting it with a name and shipping address gets it.
|
|
#
?
Aug 26, 2022 19:57
|
|
|
These rumors back up the kind of MT perf gain that Robert Hallock suggested for the 5950X vs the 7950X (up to 40%) and suggest a really high all core freq for the 7950X.
|
|
#
?
Aug 27, 2022 04:07
|
|
|
Supposedly the Zen 4 ES couldn't complete the CPU-Z benchmark until a few days ago due to BIOS/AGESA issues. The BIOS issues sound pretty rough... but the performance looks good. This Raptor vs Zen 4 battle is going to be very intense/strange in a months time. And the lurking Vcache versions make this whole match up extra bizarre.
|
|
#
?
Aug 27, 2022 04:27
|
|
|
Does Intel have some kind of potential answer to 3d cache?
|
|
#
?
Aug 27, 2022 04:32
|
|
|
Rinkles posted:Does Intel have some kind of potential answer to 3d cache? In the consumer/gamer space, Raptor has significantly enlarged caches with lower latencies than a Vcache (according to intel). It also may be cheaper to produce, so they might be able to undercut. But intel's true chiplet fighter isn't until the gen after Raptor. Zen 5 and Meteor Lake is both companies taking BIG leaps of faith, supposedly, and both are already talking a big game. Both also have rampant rumors of delays.
|
|
#
?
Aug 27, 2022 04:41
|
|
|
My guess is that the 7950X3D will probably be the ultimate CPU of the Gen but that the 13400 and 13600/k will probably be the best value for most people.
|
|
#
?
Aug 27, 2022 04:46
|
|
|
Rinkles posted:Does Intel have some kind of potential answer to 3d cache? "Hello the government. I got drunk on monopoly money and crashed my fab into a 10nm overpass. Can I have $50 billion to buy a new one? Yeah? Cool!"
|
|
#
?
Aug 27, 2022 04:46
|
|
|
MaxxBot posted:My guess is that the 7950X3D will probably be the ultimate CPU of the Gen but that the 13400 and 13600/k will probably be the best value for most people. Leaks suggest that Raptor will also continue to keep the single core performance crown, as well. Likely with the same power use asterix though, although Zen4 TDP's are supposed to rise a bit. There haven't been many (any?) leaks about Zen4 with vcache though, the specter haunting single thread gamers the world over.
|
|
#
?
Aug 27, 2022 04:59
|
|
|
Klyith posted:"Hello the government. I got drunk on monopoly money and crashed my fab into a 10nm overpass. Can I have $50 billion to buy a new one? Yeah? Cool!" It's not just the government that Intel are turning too, they're also going halfsies with a private equity firm on the financing of new fabs: https://www.theregister.com/2022/08/23/intel_asks_private_equity_firms/
|
|
#
?
Aug 27, 2022 09:52
|
|
|
ConanTheLibrarian posted:It's not just the government that Intel are turning too, they're also going halfsies with a private equity firm on the financing of new fabs: https://www.theregister.com/2022/08/23/intel_asks_private_equity_firms/ Yeah despite my sarcasm I don't think there's anything wrong with Intel getting gov't money to build a fab, as long as they follow through on becoming a hybrid fab model like Samsung that's open to manufacture for other companies. AMD got their bag when they sold GloFlo to Abu Dhabi. I think the lesson of the 10nm debacle was that this poo poo is now so difficult that even Intel can't go it alone -- trying to steal a full generation leap over the competition means you're paying for R&D the rest of the industry gets to use.
|
|
#
?
Aug 27, 2022 14:39
|
|
|
Hey AMD is welcome to book Intel fabspace! Let’s get this duopoly in HIGH gear.
|
|
#
?
Aug 27, 2022 14:55
|
|
|
I mean at this point, are there any cutting edge fabs that aren't taking government money? Maybe not always the us government, but we're getting to the point where it just takes nation state resources to keep going.
|
|
#
?
Aug 27, 2022 14:57
|
|
|
it's a strategic resource for sure too. the US doesn't like that china can threaten production whenever it threatens Taiwan.
|
|
#
?
Aug 27, 2022 15:28
|
|
|
VorpalFish posted:I mean at this point, are there any cutting edge fabs that aren't taking government money? Maybe not always the us government, but we're getting to the point where it just takes nation state resources to keep going. How much money is TSMC getting from Taiwan? TSMC was founded partially with RoC money in the 1980s and I can see stuff about Japan and the US bribing TSMC to build new fabs in their countries, but is the RoC helping TSMC with foundry capital costs?
|
|
#
?
Aug 27, 2022 15:58
|
|
|
hobbesmaster posted:How much money is TSMC getting from Taiwan? TSMC was founded partially with RoC money in the 1980s and I can see stuff about Japan and the US bribing TSMC to build new fabs in their countries, but is the RoC helping TSMC with foundry capital costs? I think it still has most of its R&D budget covered by local subsidy, but it's subsidized by so many different governments at this point it's hard to say who's helping with what.
|
|
#
?
Aug 27, 2022 16:07
|
|
|
Makes you wonder if the enthusiasm we see regarding Taiwan's independence by the West would be the same if TSMC didn't exist It's all the eggs in one basket and an extraordinarily vulnerable sector, all it takes is Xi Jinping to give the thumbs up and that's a global catastrophe-level incident
|
|
#
?
Aug 27, 2022 22:04
|
|
|
Klyith posted:Yeah despite my sarcasm I don't think there's anything wrong with Intel getting gov't money to build a fab I do, but that ship has sailed ages ago.
|
|
#
?
Aug 27, 2022 22:06
|
|
|
Zedsdeadbaby posted:Makes you wonder if the enthusiasm we see regarding Taiwan's independence by the West would be the same if TSMC didn't exist https://en.wikipedia.org/wiki/Loss_of_China?wprov=sfti1 Without going too far into politics, despite the efforts of Nixon, Reagan and HW Bush, there still exist a fair number of powerful people that think zombie Chiang is ready to be unleashed or something. If it wasn’t this it’d be something else.
|
|
#
?
Aug 27, 2022 22:56
|
|
|
Cygni posted:Leaks suggest that Raptor will also continue to keep the single core performance crown, as well. Likely with the same power use asterix though, although Zen4 TDP's are supposed to rise a bit. There haven't been many (any?) leaks about Zen4 with vcache though, the specter haunting single thread gamers the world over. I don't think there's anything particularly credible though. Other than zen4's vcache is very similar to zen3's and it will be zen5 that has redesigned cache & a next-gen vcache.
|
|
#
?
Aug 27, 2022 23:36
|
|
|
I’ve got an ASRock X570 Taichi, and I’m looking to max it out by the end of the year. I initially paired it with a 3400g because it was cheap and I couldn’t get my hands on a GPU right away, but I am looking to have a 5950X in there by the end of September. The big question, which has probably already been answered plenty of times before, is: should I care about RAM QVL lists anymore? I don’t necessarily need to squeeze the last few % of framerates out with the fastest RAM, but 64 or 128GB is likely going to come in handy during the lifetime of this computer. Also, ASRock has a “storage QVL” list on their support page for my board. Should I pay attention to this instead of grabbing the fastest PCIE4 NVMe I can find?
|
|
#
?
Aug 28, 2022 00:01
|
|
|
It's funny how my 5600 non-X is around 2% faster and consumes 10-15W less in games than my 5600X, due to the far superior binning if going by the max effective clocks versus Vcore (1.1V vs 1.3V)
Palladium fucked around with this message at 00:16 on Aug 28, 2022 |
|
#
?
Aug 28, 2022 00:13
|
|
|
Dr. Video Games 0031 posted:I do, but that ship has sailed ages ago. heh, yeah. by normal standards of politics & corporate welfare I guess it's ok. istewart posted:The big question, which has probably already been answered plenty of times before, is: should I care about RAM QVL lists anymore? I don’t necessarily need to squeeze the last few % of framerates out with the fastest RAM, but 64 or 128GB is likely going to come in handy during the lifetime of this computer. If you want to run 4 sticks of high-capacity ram and have it work at XMP speed, then yes paying attention to QVL is a good idea. If you're ok with buying 3600 and having it perform at lower speed, than buy whatever. The big downside here is not that 3200 performance is terrible. It's that when your ram doesn't run at XMP you drop to JDEC 2666 by default, and finding something in between means loving around with auto-OC or manual overclocking. And that takes some time even if you're not trying to dial in the maximum possible OC. Auto-OC for memory is often terrible and fails under memtest, so it's not really a timesaver. tldr get QVL if at all possible. istewart posted:Also, ASRock has a “storage QVL” list on their support page for my board. Should I pay attention to this instead of grabbing the fastest PCIE4 NVMe I can find? Nah I can't imagine it makes any difference for a good quality drive. That list is probably more about denying coverage for terrible crap.
|
|
#
?
Aug 28, 2022 00:19
|
|
|
hobbesmaster posted:How much money is TSMC getting from Taiwan? TSMC was founded partially with RoC money in the 1980s and I can see stuff about Japan and the US bribing TSMC to build new fabs in their countries, but is the RoC helping TSMC with foundry capital costs? Samsung and tsmc both get very high levels of government support (samsung, whose family heir just got pardoned for bribery crimes more or less explicitly because his company is too important, is also complaining that they need more incentives and subsidies from the rok government to match the much higher level tsmc gets from the roc government). They get assistance speeding up permitting, guarantees on the resources like water they need for operating, loads of tax incentives for land, construction, and operation, municipal support to get everything hooked up quickly, and generally a lot of special treatment to waive costs and wait times. This is the nature of the industry; there used to be dozens of companies that competed to fabricate semiconductors near the leading edge, and now there's three. Two or three are kinda sorta close behind and everyone else is making power circuitry or stuff for dishwashers or whatever on old nodes. These characteristics of the industry are not purely an outcome of trade or investment policy, but in large degree reflect intrinsic challenges in the physics and engineering problems that make up semiconductor fabrication. States with firms that are still in the game can decide if they want to stay in or not. Taiwan doesn't have a choice since the rest of their economy isn't growing too good and logically doesn't have any particular comparative advantage it can leverage to stay ahead of competition in the region; tsmc's share of exports keeps going up and up. Zedsdeadbaby posted:Makes you wonder if the enthusiasm we see regarding Taiwan's independence by the West would be the same if TSMC didn't exist
|
|
#
?
Aug 28, 2022 03:10
|
|
|
Eletriarnation posted:Hell of a thin client - all it needs is a more serious GPU and they could sell it as a workstation. More like thicc client I wonder if it has some way to break out some PCIe lanes for an external GPU? They can't possibly be using all of the CPU's lanes internally.
|
|
#
?
Aug 28, 2022 03:15
|
|
|
the elite mini b550 comes with a gpu dock
|
|
#
?
Aug 28, 2022 03:25
|
|
|
Mini my butthole.
|
|
#
?
Aug 28, 2022 03:27
|
|
|
i spent the past couple days digging into if there's a mini pc that'd be better than my current "thin" client solution which is using my macbook air. it seems like there's still no real competition for apple silicon in terms of performance/watt right now
|
|
#
?
Aug 28, 2022 03:32
|
|
|
redeyes posted:Mini my butthole. Reminds me of ITX cases that is only 7L smaller than a 37L ATX one
|
|
#
?
Aug 28, 2022 03:43
|
|
|
The extra psu and full length graphics card make it real goofy. If they had a flex-atx psu and a short graphics card it would look like a weird cross between an open case, a nuc, and a truly small case like a velka or whatever, and that would be kinda neat if silly. But with a sfx psu and full sized graphics card it feels more like a cross between an open case, a nuc, and those external gpu enclosures for laptops that were around for a few years.
|
|
#
?
Aug 28, 2022 03:55
|
|
|
|
| # ? Jun 6, 2024 08:57 |
|
|
with USB4 on the incoming AMD chips, hopefully there'll be a resurgence in egpu solutions i have two razer core x sitting in my closet unused just because of how much of a PITA using an egpu on AMD has been even with a thunderbolt daughter card
|
|
#
?
Aug 28, 2022 04:09
|
|