/v/ - Video Games

Vidya Gaems

Index Catalog Archive Bottom Refresh
Options
Subject
Message

Max message length: 8001

Files

Max file size: 32.00 MB

Max files: 5

Supported file types: GIF, JPG, PNG, WebM, OGG, and more

E-mail
Password

(used to delete files and postings)

Misc

Remember to follow the Rules

The backup domain is located at 8chan.se. .cc is a third fallback. TOR access can be found here, or you can access the TOR portal from the clearnet at Redchannit 2.0.

Maintenance complete. Other systems remain to be tested and updates will come as they are ready.
Ghost Screen: Especial de Halloween primer sábado
Comienza a las 0UTC
El Hilo


8chan.moe is a hobby project with no affiliation whatsoever to the administration of any other "8chan" site, past or present.

Reminder that 8chan.se exists, and feel free to check out our friends at: Animanga ES

PC Hardware Thread + News: Battlestation Edition Anonymous 09/01/2022 (Thu) 15:56:48 Id: 104321 No. 680781
Share your build specs to rate and rig pictures if you'd like. Then others can rate. It's news time, no nvidia stuff this time around since there's no solid info. AMD Ryzen 7950X, 7900X, 7700X, & 7600X Specs, Price, Release Date https://yewtu.be/hVnJbiYOCq4 AMD Ryzen 7000 desktop chips based on Zen 4 architecture coming September 27th for $299 and up >AMD says when compared with a previous-gen Ryzen 9 5950X processor, the new Ryzen 9 7950X chip delivers <Up to 29% better single-core performance <Up to 45% “more compute for content creators in POV Ray <Up to 15% “faster gaming performance in select titles” https://archive.ph/spgRW >During its event, AMD made some bold performance claims versus Intel’s 12th Gen Core processors based on Alder Lake. First and foremost, AMD claims the entire Ryzen 7000 series line-up, including the lower-power, more mainstream Ryzen 7 7700X and Ryzen 5 7600X, will offer better single-thread performance than Intel’s top-end Core i9-12900K. >To that end, a Ryzen 5 7600X was shown running F1 2022 side-by-side with a 12900K, and the Ryzen put up an average framerate approximately 11% higher, though both chips were capable of hitting north of 350 FPS in the test. >On average, AMD is claiming about 5% better gaming performance with the Ryzen 7000 series versus Intel with the 7600X, with faster parts obviously increasing that lead. https://archive.ph/fhSAU AMD Radeon RX 7000 series: Everything we know about the RDNA 3 GPU >Recent rumors suggested that RX 7000 graphics cards could be up to 130% faster than the previous generation overall while staying relatively power-efficient. While AMD is still quiet about the details, it has confirmed that the new GPU series will offer more than a 50% improvement in performance per watt over the previous generation. We also have confirmation that AMD will be adopting the 5nm manufacturing process, which should help drive some of this performance boost. >AMD has confirmed it will make use of a new chiplet design borrowed from the architecture used on its Ryzen processors for RDNA 3. This change in design could allow the company to double the number of compute units on each die. If all else is the same, we can expect AMD’s GPUs to ship with at least 12GB of VRAM. The company had previously claimed that 12GB of RAM on GPUs is the minimum requirement to be future-proof when playing AAA titles. Current AMD GPUs ship with GDDR6 memory, and we can expect the same with Radeon RX 7000 since Nvidia has an exclusive on the faster GDDR6X standard. https://archive.ph/ZSCTT Previous Bread: https://archive.ph/stG4B
>>680792 >I kinda wanna upgrade from the best DDR4 platform to the worst DDR5 platform
>>701329 >the really shit boards struggle after 5200 when they really shouldn't and JEDEC goes all the way up to 6400. That's the CPU's weakness, not the motherboard's. DDR5 IMCs are still immature.
>>701331 >do JEDEC vs XMP matter in real life usage? It matters if your motherboard does not support XMP, pic related
>>701331 >do JEDEC vs XMP matter in real life usage? it would suck to buy a DDR4 3600 XMP kit, but your mobo doesn't have RAM overclocking, so it runs at 2133MHz instead. Buying a 3200MHz JEDEC kit from the start would avoid that.
>>702648 Only 12500 and down really have that issue and they still don't struggle that low on a proper board. 12600 and up all should do 6400 JEDEC or better. Right now it's mobo and ICs that are the biggest factor, considering even mediocre chips will do 7200+ with hynix A die on a great board.
>>702658 >12600 and up all should do 6400 JEDEC or better. Without manual overclocking or XMP support, the CPU and RAM autonegotiate to the greatest common denominator between what the CPU supports and what JEDEC profiles are programmed to the RAM's SPD. Alder Lake only officially supports up to DDR5 4800. Anything beyond that is overclocking. While you are correct that most CPUs can extend beyond their official specs, that is not what's being discussed here. For another real-world example, if you stick a JEDEC DDR4 3200MHz kit in an X99 board with a 6950X, the RAM will only run at 2400MHz unless you enable XMP or manually overclock. Because Broadwell-E only officially supports up to 2400MHz. Yet with overclocking you can indeed go past 2666MHz. I know this from personal hands-on experience with those parts.
>>702658 >>702738 Or another example, my girlfrien'd B360 board has a 2666MHz kit installed, but her i3-8100 CPU only supports 2400MHz, so that's what the kit runs at. If she had a motherboard that supported memory overclocking then the SPD speed becomes irrelevant
The Dell XPS 8950 can't even run four DIMMs at 4800MHz
>>702658 >Right now it's mobo and ICs that are the biggest factor I agree with you now, sorry for my multiple posting, getting sleepy and sloppy
>>702738 >>702739 >that is not what's being discussed here. True but there's a grand total of 0 reasons to stick with JEDEC unless your hardware actual locks out the ability to do so and even then it's almost always worth changing it to something that cna. >Because Broadwell-E only officially supports up to 2400MHz. Well that and the IMC is a dumpster fire with some ICs/PCB designs. >>702744 4 sticks on intel DDR5 is literal fucking death, unless you have an ASUS board, then some of those can do 6000-6400
>>702749 >unless your hardware actual locks out the ability to do so right, the context is my family's Dell and HP prebuilt office PCs, they almost never support overclocking
>>688142 I got my new PC into one of these this year. It's pretty cool.
>>702749 >4 sticks on intel DDR5 is literal fucking death So RAM watercooling isn't a meme?
>>703444 Nah it's the RAM training part (aka where the mobo talks to the RAM for a while to get what it wants and supports), ASUS has it down so you can expect decent speeds (6000-6400), for now everyone else you hard cap at 4800-5200 DDR5 is not that hot unless you really push it (1.6v+).
>>703469 How long until we get 10000+ RAM speeds? I'm hoping it happens during the Ryzen 9950X generation.
>>703475 I'd say 3-5 years, less if you're fine with extreme daily OC, technically already done on LN2. Can't guarantee it'll be actually good or worth doing, since platforms can have a frequency sweet spot (like say Zen 3 being around 3800-4200 which is very far from the best you can achieve on great DDR4 sticks) lower than max speed and you can have ICs made for pure speed but are trash at everything else (Hynix DJR, can do 6000+ somewhat easily, will get clapped in performance by 4400-4800 Samsung Bdie and some halfway decent timings).
(535.53 KB 1024x520 Just fine fire.jpg)

Was the FX-9000 series with their 200W+ TDPs ahead of the time?
Performance difference between box cooler and high-liquid cooling of 7600x is out, and it's just 5%. You get barely more performance out while using better cooler. Even using the ECO mode with highest choke for the supposed lower system power you net usable performance (same as 5800x3d in r23). The MAIN difference you get is temps, which again goes from 77C best case to 101C worst case. Hence the "95C is normal now". Single core performance difference is negligible, so is gaming. AMD could've just released 7000s with lower thermal output and still have a great product, guess they didn't want any chances with 13th gen and went all out, was this the idea.
>>704007 Just watched this myself. https://invidious.snopyta.org/watch?v=AiMcQB2FvyM So if you are forced to use a small CPU air cooler you can do that, but might break 100C. Best bet is to just take the small performance hit and enable Eco mode if your cooling is limited since you don't want to be dumping >95C of heat into a small enclosure in which components are close together. A standard sized enclosure with limited fans would also benefit. Eco mode shows temps everyone is more accustomed to while Stock is basically "We can do that too!" since Intel has been hamming up the clocks, power consumption and heat output saying "Fuck efficiency!" for a long while now.
>>703998 All depends on the die size, bigger die means more wattage is easy to dissipate, Bulldozer is insanely large for a CPU so even 200W on air is not impossible, Zen4 is ridiculously small so even 150W is starting to not be coolable, and that's not even accounting for Bulldozer needing to stick under 70°C whereas Zen4 might handle 100°C with a similar level of degradation to the chips. >>704007 I mean Pentium 4 on stock coolers were already riding 100°C for some configurations, though intel 90nm/65nm is fucking unkillable.
(77.08 KB 421x449 behold.jpg)

How do you like my boomer CPU? Til death do us part.
>>704021 >>704026 Yeah we're back in the territory of 90C under load is back normal. Would be funny to see air coolers trying their best to keep it under 80, maybe some thermal caps on OEM el cheapo Lenovo prebuilts. Keep in mind this doesn't show sustained load or throttling in charts. Let's see how a 80C thermal limit nets a decent performance
>>704046 How is amd's fine wine technology holding up? Would be good to see some 6300 or athlons I know you fags are still running those
>>704057 I do have an undervolted passive build with a Phenom 2 X4 955 for basic internet browsing, which itself replaced a 6400X2 build that died from my own retardation never put locking SATA cable in vertical headers > Let's see how a 80C thermal limit nets a decent performance There was a few tests where capping to 85°C netted better perf, which is logical because higher temps cause leakage which cause higher temps which causes more leakage etc
So intel arcs flagship is launching this month and is supposed to be around a 3060 in performance. But I can't find any mention of anyone testing the lower end GPUs for virtual machine or vfio type shit. No mention of whether or not GVT-g works or anything. Seems kind of odd since this would be rather significant to test and discuss.
>>704026 Are we getting to the point where we're going to have to start installing small consumer-grade liquid nitrogen condensers next to our PCs so that they can drop-feed directly through the copper pipes? I can't run a condenser and my GPU on the same circuit without blowing a breaker
>>704359 Doesn't delidding get a massive amount of cooling results? That's been a thing for decades now.
>>704374 I was just looking up an old Shills's Hardware post on that, and I think they were getting a roughly 10 degree decrease on average by delidding, which isn't massive - but 10 degrees is 10 degrees, and opens up some more headroom to overclock. They also just added better thermal paste, then replaced the lid. They didn't fuck around with directly cooling the die.
>>702639 The worst DDR5 is still better than the best DDR4.
>>704327 Arc doesn't get see lot of testing because the drivers are awful and no one wants to invest time grappling with that when they can be doing other content.
>>704359 Phase change is so much easier to do and so much less power hungry Especially if you just near near ambient temps and not sub zero. >>704394 Not even remotely fucking close The worst DDR5 spec is slightly worse than the best JEDEC DDR4 (both 3200 CL22 but DDR5 has worse secondaries and tertiaries) You only start having a clear advantage with good DDR4 vs best current DDR5 in practical workload (aka not AIDA64 bandwidth test) on, ce you hit well tightened 6800 And that's not even considering the abomination that are half rank DDR5 sticks which is what all the cheap shit uses.
>>704046 To my knowledge, that's basically the newest and fastest non-pozzed x86 platform
>>704394 >The worst DDR5 is still better than the best DDR4. That's like saying a 2022 Chevy Spark is better than a 2012 Corvette
>Newegg is bundling DDR4 RAM with DDR5 motherboards Sometimes you can admire a well-pulled-off scam. This isn't one of those, and is just retarded kikery.
(1.70 MB 400x225 this is fine 3D.gif)

>>704818 At least they're not bundling exploding Power Supplies that will burn your house down.
>>704831 Not anymore, gigabyte ran out of exploding PSU.


Forms
Delete
Report
Quick Reply