>7% performance gain over the 9950x>costs 80% more>4% performance gain over the 9950x3d>costs 40% moreuse case for the AMD Ryzen 9950X3D2?
>>108653332Call of Duty
What, you thought the death of Moore's Law was just a meme?
>>108653332I'll justify it to myself somehow
It's funny because this is the standard in every other industry.A car having 7% faster top speed is usually 3-10x more expensive.Somehow we got so used to regular gains that we now consider such a small price difference for that extra performance ridiculous. This is how all mature technologies work.
https://www.techpowerup.com/348400/no-ryzen-9950x3d2-for-techpowerup-gamers-nexus-or-computerbaseLOLMAO AYYMDEAD KNOW IT'S SHIT SO THEY DON'T LET REVIEWERS REVIEW IT
It's the Pentium 4 Extreme Edition or AMD Athlon 64 X2 FX-62 of today. A meme overpriced cpu for people who can't overclock.
>>108653332Lisa Su needs more gibs for lobbyistsShe's not Jensen they don't pick up her calls if she doesn't turn up with bags of bribes
>>108653594That's not the problem. The problem is that all the performance increases so far have been theoretical for the most part. There is so much software from twenty years ago that still performs like shit.
>>108653332Apparently you still need to core park with this. I thought the point was you’d get slightly less performance than the 9950x3d but not have to deal with all that bullshit.
>>108653616>>Apparently you still need to core park with this.Whatplease be lyingWhy???
>>108653624NTA, because:>OS schedulers really like moving processes around>and moving processes invalidates all sorts of caches>which then have to be repopulated from memory>and you can schedule about 2000 ADD instructions in the time you wait for 64 bytes of memory
>>108653624cross ccd latencyremember your zen 2? you had 2 ccx on 1 die, if you had a 3600 you effectively had 2 tri-core cpus in your system
7-10% faster is the whole reason stupid nigs pay $300-500 extra for nvidia gpu over amd equivalent
>>108653704lol.
>>108653704I don't think the people using computer parts as fashion accessories can grasp percentages
this is what you get for not buying and supporting risc-vyou deserve this bullshit and even worse
>>108653332it probably adds some instruction or extra pci lanes or wider bus or something that has its niche uses across various industries.
>>108653714The one at the bottom reminds me of overtuned post processing/enb settings
>>108653332the use case is getting stupid people to stfu about adding the cache to both ccds expecting more performance
>>108653332CFDs and other super nerd stuff
They should have just made the vcache span both CCDs instead of each having it individually...
>>108653332there's like 5 tests it does over performbut you really have to spend most of your time on that list to consider buying it
this is what happens when they keep releasing the same chip reheated for nearly 2 years straight x86 is so embarrassing. why can't they ship new parts regularly
>>108653332mogging the brokies
>>108653332this is a chip for farming whalesit's so stupid i thought /g/tards would like it
>AMD says it's fucking retarded to make a 9950x3d2, it won't make significant difference and it'll way more expensive>"NOOOOO YOU GUYS NEED TO MAKE ONE, WE WANT TO CONSOOM">AMD releases the retarded chip because they have nothing better to do>retards complain about the chip being mediocre for the pricejust lol and lmao
>>108653332Use case for Intel Extreme black label CPUs since Core2Duo/Quad days?
>>108655852AMD could have solved this easily. they have an entire series of epyc AM5 SKUs which never had reviewsif they had launched it there the price would even make sensebut they wanted more moneythen the retards and poor fluid dynamic simulators who wanted their dual cache variant could have it without complaining
>>108653332AMD told people for years that adding more cache to the second CCD would do almost nothing. But people bitched and moaned and complained about wanting it, so finally AMD gave in and released one... at which point those same people said they thought it was pointless all along and a complete waste of time.The moral is you should never listen to your customers, because most of your customers are fucking retards who don't even know what they want.
>>108653332>use case for the AMD Ryzen 9950X3D2?eliminating CPU bottleneck for RTX PRO 6000 owners
>>108653714lol. this you bruv?
>>108656094https://desuarchive.org/g/search/image/yTl7BDkdkucIquYcXUSYlw/There's no doubt it's all the same guy.
>>108653714WTF I love Linux now
>>108656126Oof, one of the generalbots as well. AMDrones are pathetic.
>>108656126Whoops, cut off the most important part of that image too.
poorfag intelkikes seething and can release the most retarded SKUs that no one cares about and still no one will Bulldozer 2 (AKA Arrow Lake)
I'm just trying to work out who its for. As far as the reviews I've read it mostly is best for compiling code.
>>108653332>use case for the AMD Ryzen 9950X3D2?none, they should make a 4channel hedt if they want people to be able to actually benefit from so many fast cores,I cna guarantee you that these cpu are basically spending 90% of their time waiting
>>108653332None, they said in the past there was no difference but people still bitched and asked for dual V-cache. There it is your halo product.
>>108656210it's like 3 seconds faster to compile the linux kernelwho is it for?
>>108656298
>>108653594But if that's the case there are CPUs for that. This is just regular desktop stuff
>>108656282AMD fanboys getting what they deserve? As foretold in the dead seal scrolls.
>>108653332it was made by amd to prove a point about how wasteful dual 3d v-cache isthey'll never do it again
>>108656210it's for people who like to brag on forums instead of actually playing games>>108656298>>108656307linus uses threadripper not this pleb consumer cpu
>>108653332Yeah not for me, I'm still sticking to my 7950X3D for now.
>>108653332Yeah not for me, I'm still sticking to my 5800X3D for now.
>>108656509>it's for people who like to brag on forums instead of actually playing gamesHey, I like playing games!Especially those where you get to kill cops for shits and giggles.
>>108656518>>108656558What will the 11950X3D be like
>>10865660724 cores with 12 cores per ccd
if you want an actually unified huge LLC you will need to wait until Nova Lake shows AMD how to make a cache
>>108653332None. Should've used that silicon to rerun the 5800X3D.
>>108653332>>108653596>>108653704>>108655764>>108656563>>108656674It performs WORSE in games than the 9850X3D half its price lmao>wccftech.com/amds-900-ryzen-9-9950x3d2-dual-edition-flops-in-reviews-losing-to-a-chip-half-its-price-in-gaming/
Let's see the power usage numbers.
>>108656816What is the power level gap between 5800X3D, 7800X3D, 9800X3D and 9850X3D
>>108656853Linux kernel compile times (lower is better):>AMD Ryzen 7 9850X3D 8-Core 1044>AMD Ryzen 7 9800X3D 8-Core 1111 +6%>AMD Ryzen 7 7800X3D 8-Core 1381+32%>AMD Ryzen 7 5800X3D 8-Core 2035+95%
>>108653332>7% performance gain over the 9950xThat's in the best case, actually it loses to 9950X in production tasks that care little for cache due to clockspeed difference. It cannot boost as high and margains are much tighter in most production stuff as cache is not some magical panacea in productivity and clockspeed tends to matter more for said tasks. >>4% performance gain over the 9950x3dAgain best case, in practice 9950X3D is slightly faster in productivity that uses more then 8 threads for similar reasons as the non-Vcache CCD can booster higher. >use case for the AMD Ryzen 9950X3D2There's a reason why AMD never made a dual-CCD V-cache Ryzen SKU until now. I suspect this release is just pure bragging rights meant to milk /v/tard whales who don't know any better. For pure pearless gayming, just stick with 9800X3D or 9850X3D for AM5.
>>108655441You would get a proper Epyc or Threadrippoer build (getting the cache heavy SKUs) for that. X3DX2 only makes sense if you need the cache but cannot afford or justify the Epyc/Threadripper platform.
>>108656044Extreme Edition SKUs and Black Edition back in hte day were were fully unlock chips that usually came from good bins.These are just excessive V-cache yields that didn't make the Epyc/Threadripper cut and 9800X3D/9850X3D units weren't moving fast enough.
>>108656905meme benchmark, barely no difference between the X3D and non X3D chips
>>108656816It performs the same, that's just margin of error. AMD said a dozen of times it would perform the same but people insisted they need dual cache
>>108656853
>>108657045>12 game averageHow do they measure this exactly? Every game uses a different engine and has different FPS limits
>>108657123>fps limitwhat fucking modern popular game is locked by an FPS cap?
>>108657181>what fucking modern popular game is locked by an FPS cap?Many engines are built for 60fps, 120fps or 240fps
>>108653332If you care about the price it's not a product aimed to you
>>108657037>dual cacheQRD
>>108657205it has 5 chipsiod, ccd0 and ccd1both ccd have an extra cache chip addedAMD didn't bother previously because it was actually detrimental to clock speed but with 9000 series they decided to show everyone how little it does even when clocked high
>>108657123They pick 12 games and run the same scenes. The goal is to simply get deltas.
>>108657187yeah maybe over a decade ago, nowadays basically anything can run at 1000fps+ if you got the power, and unless you are capcom, nothing will break
>>108657205It performs well across the board with no performance downsides unlike the assymetric 7950x3d and 9950x3dIt does use slightly more power and cost way too much though.
>>108657192Not everyone defecates money, pretentious worm
>>108656918>7%>That's in the best caseSasuga Windows. On Linux, it averages 10% faster than both 9950x and 9950x3d.
>>108659634Linux is 0.20% marketshate
>>108659634All these multi die chips have always run better on linux than windows, because the algorithm was implemented before, but I've seen reviews of this and in the best case scenario it had less performance than a 9800x3d.
>>108653332what makes you think performance gain is a metric?
>>108661080Is that number from the Steam Chinese internet cafe hardware survey? Either way, I only care about its market share on my desk (100%).
>>108661080Windows doesn’t really matter when it comes to computers It’s a ranch for goy cattle
>>108653332Daily reminder that x3d only excels because amd uses a horrible io controller
>>108653594Computers and cars couldn’t be more similar these days. The Neo will be the Toyota Corolla of laptops for the next 5 years.
>>108653332> 4% performance gain over the 9950x3dIn gaming? I think it's worse due to lower clocks.
>>108656281Yes, it already had existed. You have to pay the Threadripper Pro/Epyc platform tax for it.
>>108659634Kiddo, their test suite uses a good number of applications/test that actually takes advantage of extra cache and threads which makes 9950X3D2 more rosy then it actually is in practice for most applications under *nix ecosystem.Again, this product only make sense if your use genuinely takes advantage of the threads and cache but you cannot afford or justify the Epyc/Threadripper Pro platform tax.
>>108661208Wrong, it exists because Ryzen ecosystem are basically a multi-socket platform on a chip. Intel is doing the same thing as they made the move to chiplets (starting with Arrow Lake). This is why Arrow Lake got a cache bump under "Plus refresh" and upcoming Nova Lake is going heavy on L3 cache.
>>108661080in china
>>1086568169950X3D still the best deal for gaming. I wouldn't want to wait double as long for game installation and shader compilation to finish.
>>108661442Literal fucking bugmen
>>108661554No, the best deal for gayming is actually Intel. For min-maxing without going full pants-on-head retard, it would be 9800X3D. 9950X3D only make sense if you want to do productivity on the side.
>>108653332Better win10 performance without having to lasso your tasks to the x3d ccd threads.
>>108656931>X3DX2 only makes sense if you need the cache but cannot afford or justify the Epyc/Threadripper platformexactly
>>108653332I get all off mine CPU for free (9950x3d2 as well) for free due to reasons so use case for me is:> COST:0>PERFORMANCE:4% over others.>Free 4 %I don't actually use it but that's the logic.
>>108656931That's a great deal.
gaymers
>>108653594>car analogycan't make this shit up
>>108656931x3d TRs exist? i thought they were saving all that silicon for epyc
but they still outperform intel
>>108656210>As far as the reviews I've read it mostly is best for compiling code.Extremely useful for AI.iykyk
>>108658075Do you get mad at expensive gold watches? I don't, I wouldn't ever buy one
>>108653714Now try cyberpunk or CS2, lil labubu unc
>>108653332just so AMD can claim they have the fastest CPUsame as 13900KYS/14900KYS a couple years ago
!!! DO NOT BUY !!!
>>108667257see, amd sent him a review sample afterall
I don't expect this shit to get any better soon, they are hitting the limit on current silicon technology. If intel doesn't fuck up then they will be better in ~3-5 years because they are making new fabs with new tech that will take TSMC to catch up to... Unless intel fucks up. You had a good run AMD, now stop fucking us
Wake the fuck upAMD never really intended to make this shit, but with the upcoming dual bLLC Intel CPUs, 9950X3D2's sole purpose is so that AMD can claim they were the first ever to ever release dual big L3 cache in a CPUIt was never supposed to be good value, move on
>>108667627we will get one last hurrah with 2nm+GAA+BSPD and everything after that will be single digit improvements every 3 years for 50% higher cost
>AMD is showing the first signs of turning into Intel>Intel got so buttfucked they admitted to not being able to catch up any time soon and fucked offfuck
It's a flagship device that's expensive to make. The fuck do you expect?
>>108667756Surely nvidia will save the consumer processor space soon!
>>108667756its already set in stone that novalake will blow zen6 out of the water. the 9950x3d2 is amd's last time to practice kesef kal
>>108653332>ask for product >amd says it's pointless>keep asking >they give you what you want>cry about itconsumers are retarded cattle
>>108667833its a huge cash grab. there's no reason why having adding another 64mb of cache on top of the 9950x3d would cost $200-250 more. the actual performance of the product is good though
>>108668011it's normal pricing for a HEDT part
>>108668114its priced at amd's opportunity cost, i.e. the cost of two 9800x3d's
>>108668011>there's no reason why having adding another 64mb of cache on top of the 9950x3d would cost $200-250 more. Wtf are kids being taught in economics classes these days?
>>108668011>the actual performance of the product is good thoughThe majority of the people buying this part aren't even considering any of that. They're just picking up the top sku. This part is marketed towards people that are just grabbing things off the top shelf. Just like the 9950x3d.
>>108668114well its not hedt. it has a 128bit memory bus, 20 pcie lanes and 4 chipset lanes. and not even ecc is supported officially
>>108668222all ddr5 is ecc
>>108668374on-die ecc isnt the same as traditional ecc
