AMD Launches Lower Cost 12- and 24-Core 2nd Gen Ryzen Threadripper Chips (hothardware.com)

MojoKid writes: AMD launched its line of second generation Ryzen Threadripper CPUs over the summer, but the company offered 16-core and 32-core versions of it only at the time. Today however, the company began shipping 12-core and 24-core versions of the high-end desktop and workstation chips, dubbed Ryzen Threadripper 2920X and 2970WX, respectively. All 2nd Generation Ryzen Threadripper processors feature an enhanced boost algorithm that came with AMD's Zen+ architecture that is more opportunistic and can boost more cores, more often. They also offer higher-clocks, lower-latency, and are somewhat more tolerant of higher memory speeds. All of AMD's Ryzen Threadripper processors feature 512K of L2 cache per core (6MB total on the 2920X and 12MB on the 2970WX), quad-channel memory controllers (2+2), and are outfitted with 64 integrated PCI Express Gen 3 lanes. The new Ryzen Threadripper 2920X has a 180W TDP, while the 2970WX has a beefier 250W TDP. In highly threaded workloads, the Threadripper 2920X outpaces a far more expensive 10-core Intel Core i9-7900X, while the 24-core / 48-thread Threadripper 2970WX is the second most powerful desktop processor money can buy right now. It's faster than Intel's flagship Core i9-7980XE, and trailed only AMD's own 32-core Threadripper 2990WX. Pricing for the new chips falls in at $649 for the 12-core 2920X and $1299 for the 24-core Threadripper 2970WX.

Comments

Hello intel my old friend

[Anonymous Coward]

2920X = $649
i9-9900K = $580

Frankly you'd be nuts to go for the intel chip when you can get 50% more cores AND an upgrade path to 64 cores for around the same price.

Re:Hello intel my old friend

[Anonymous Coward]

This reminds me how Edison developed the electric chair to prove AC was dangerous!

Re: Hello intel my old friend

[Anonymous Coward]

A video from 2014. Great work fanboy. Can't you fins anything newer?

Anyway, with Intel there's no money left to buy the eggs. Hehe.

Re:Hello intel my old friend

[corydoras]

But the single core performance seemed underwhelming on the new AMD processors, especially for the price. It seems like virtually everything I do is limited by single thread performance, with multiple cores mostly being good for multitasking. Am I missing something?

Re:Hello intel my old friend

[corydoras]

It seems for instance I could get a Pentium G5600 and outperform the AMD 1950X in most cases, while saving a ton of money.

Re:Hello intel my old friend

[Anonymous Coward]

But the single core performance seemed underwhelming on the new AMD processors, especially for the price. It seems like virtually everything I do is limited by single thread performance, with multiple cores mostly being good for multitasking.

Am I missing something?

Maybe, because virtually everything I do is multi threaded. Anyway, the single core performance is actually very close to Intel's processors, even if you are using something single threaded.

Re:Hello intel my old friend

[Anonymous Coward]

If you only care about single core performance, you could buy a lowend chip and be just as happy. Intel actually hasn't made much improvement with that aside from the newest 9th gen chips and those are hard to find. A ryzen 7 2700x is as fast as most core i7 chips from the last few years with the exception of 8th and 9th gen.

you are not the target audience for this chip. it's not for gaming. It's not for running old single threaded apps.

Re:Hello intel my old friend

[Anonymous Coward]

That son of a bitch actually tortured and electrocuted a live Elephant in his demonstrations.

That's your big hero Edison.

VWORD: sinister

Re:Hello intel my old friend

[Anonymous Coward]

Depends on your use case..

I have a Ryzen 1600 (6 core @ 3.2Ghz base clock) and an intel G5500 (2 core @ 3.8ghz base clock), for dolphin emulator which really only makes use of 2 cores and predominately relies on better single core performance the low end intel chip wins out in raw performance (not to mention price performance consideration). Now if I am trans-coding a blueray to h.265 the software makes use of all the cores of the ryzen, so in that use the g5500 is absolute garbage in comparison.

Right chip for job..

Re:Hello intel my old friend

[K. S. Kyosuke]

They removed that feature, sadly.

Re:Hello intel my old friend

[hairyfeet]

Maybe...if you are running Win98. Today's OSes are more than capable of multitasking so even if you are just doing nothing but simple browsing those extra threads can come in handy, for example just browsing I also have a temp monitor, AMD Customer Experience (reports if a game has bugs or crashes so they can release patches), several threads for the browser, Steam checking for updates to my games, etc.

So even with my older FX-8320e I've found the extra threads frankly more useful than more single thread performance as I don't have to think "do I have enough oomph for this?", if I suddenly decide to watch a little classic Vincent Price for Halloween (Theatre Of Blood 1973, love 70s Price) while I'm rendering some video? No problem, I just do it. If my AV wants to update itself while I'm gaming? Don't care, never notice. Nothing skips, nothing stutters, The PC can happily do its background tasks even if I'm chopping baddies to sushi in Shadow Warrior 2 or letting go with a full broadsides in World Of Warships and I'll never be bothered,it all "just works".

You try doing that on that G5600 and you are gonna be stuttering so bad you'll think you are on a P4, so if that makes ya happy go for it, but frankly I'd take even a Ryzen 3 over the G5600 any day of the week which just FYI is also $10 cheaper on Newegg than the 5600.

Re:Hello intel my old friend

[Anonymous Coward]

Frankly, you'd be nuts to spend $650 on a CPU.

Re:Hello intel my old friend

[Tough Love]

Reports are that the 9900K draws way more than 95 watts when running overclocked for the fiddled benchmarks. A lot of complaints about cooling problems out there. A lot of doubt about accuracy of benchmarks. And the chip is out of stock everywhere, so a lot of people are calling it a paper release. A lot of talk about cancelling orders and going with 2700X or Threadripper instead.

Re:Hello intel my old friend

[Tough Love]

the single core performance seemed underwhelming on the new AMD processors, especially for the price.

Especially for the price? You've got to be kidding. The 8 core 2700 sells for $265 right now, 6 core 2600 for $160. And single core performance is respectable, I have no complaints at all. Multicore smoothness is great even if you aren't running compiles for a living. You never get some out of control web page slowing down everything the way it used to be. Mind you, I'm looking forward to the Zen 2 announce, less than 3 months from now. Most likely equivalent IPC to Intel parts while soundly beating them by every other measure.

Re: Hello intel my old friend

[Anonymous Coward]

This is why I spent 2k on a cpu. Not everyone just plays solitaire.

Re:Hello intel my old friend

[Anonymous Coward]

I work in a lab that has various Intel processors. I thought I was all clever by getting the somewhat lower frequency machine with a lot of cores, thinking I would parallel everyone else to death. In the end, I discarded that 24 virtual core machine for an 8 virtual core that ran about 1.5 ghz faster per thread. The 8 virtual core was that much better for everyday desktop use, and when I did time runs on parallel apps, they were nearly equivalent.

I learned a hard lesson from that, and since then just focus on the biggest ghz with lower number of cores, and big fat DDR chips. And SSDs of course.

Re:Hello intel my old friend

[thegarbz]

If you're running true single core workloads then you're missing the core boost advantage of Zen+. Yeah at stock frequency they are outpaced by Intel. However when I look at my own Zen+ Ryzen if I run single core loads I typically see 300MHz over a loaded multi-threadded situation on that core, and over 600MHz above the base clock rate. This is also the reason why overclockers pushing the hell out of the chip don't see much single threadded improvement in their benchmarks.

In terms of peformance vs Intel they are only marginaly behind. In terms of performance over the industry over time well that is a depressing result for the entire industry as a whole when you compare the past 5 years. For instance my recent upgrade from a 5 year old i5 netted me a 20% improvement for single threading. Had I gone intel I would have had a 24% improvment! Hurrah! Mind you I got a 200% increase on multi-threadded workloads though.

Speaking of multithreading, full disclosure:
Posted using Firefox with a single tab open currently running more than 6 threads.

The world isn't as single threadded as you may think.

Re: Hello intel my old friend

[Anonymous Coward]

Not everyone is a complete dipshit like you, either.

Re:Hello intel my old friend

[spth]

German magazine c't did some measurements in their current 23rd volume, pages 100 to 102.

Running AVX2 code, the 9900K draws 148 W for up to 28 seconds, 50% over TDP.

The 28 seconds are just enough to complete a run through multithreaded Cinebench.

Previous Intel processors only went 25% over TDP for up to 8 seconds.

Re:Hello intel my old friend

[Anonymous Coward]

"Running AVX2 code, the 9900K draws 148 W for up to 28 seconds, 50% over TDP."

That is meaningless. Intel TDP is for base clocks only.
Look at the cooling recommendation (130W) for a better idea of medium term power usage.
It can burst above this for short periods.

Re: Hello intel my old friend

[Anonymous Coward]

No shit about the ninth gen chips; I canâ(TM)t find Coffee Lake Xeons for sale anywhere. I was wanting to assemble a C246 based workstation, but now Iâ(TM)m looking at AMD.

Re: Hello intel my old friend

[Anonymous Coward]

Or maybe some of us have disposable income and dont have issue spending it. Maybe try not being a broke waste of space living off ramen and cardboard box furniture.

Re:Hello intel my old friend

[Anonymous Coward]

Your first mistake was not taking into account your workflow throughput. If you have even a single application in that pipeline that single threads and processes in serial, your whole pipeline will bottleneck. If you have access to the source code, find that single instance, create a queue for its work and async its operation.

Low speed, high core count machines are perfect for these kind of processing pipelines or database server work, where the code optimizes for available processors. Put a bunch of single threaded pipelines and you will be sorely disappointed by your lackluster results.

Re:Hello intel my old friend

[Anonymous Coward]

Lots of bitching online about buying a new motherboard while waiting on the chip. It's called the "bleeding edge" for a reason, and those folks got cut.

Re:Hello intel my old friend

[DigiShaman]

Modern OS kernels handle multi-threading better too. Because it can allocate threads to the most free cores, it substantially reduces latency. And as you've pointed out, a far more responsive user environment and experience.

There is a balance however. At a certain point (for your average user workload), there's no need to continue pouring money in additional cores. After having 4 to 8 cores, the extra money would be better thrown towards higher clock rates. For example, I'd much rather have the fastest clocked i5 over the slowest clocked i7. That's because many problems can only be calculated sequentially (non multi-threaded), and a faster clock rate will burn through those problems much faster.

Re: Hello intel my old friend

[Anonymous Coward]

Well I bought mine for work so the 2k was well spent. I guess you dislike people using technology for making money.

Re:Hello intel my old friend

[drinkypoo]

After having 4 to 8 cores, the extra money would be better thrown towards higher clock rates.

If you're a gamer, that's certainly true. But most heavy tasks are highly parallel today, and if they aren't, they will be tomorrow. If you're doing anything that requires heavy lifting, odds are good that it will benefit more from more cores than from a small increase in single-thread performance.

Re: Hello intel my old friend

[LordWabbit2]

"am I missing something?"
Yes, multi threaded software, we ALWAYS code for multiple CPU's.

Re: Hello intel my old friend

[Anonymous Coward]

I just get all tingly when people can get together and agree to disagree in a calm, non-judgmental civilized manner.

Re:Hello intel my old friend

[Anonymous Coward]

I am running Win98.
It is *VERY* fast on the 2920X, much faster than the old P5 it replaced. I no longer have any problems waiting or running out of memory when loading my huge MSPaint projects, Office works fine, and Internet Explorer is simply dreamy.
It also runs OpenBSD.
I highly recommend AMD products.

Re:Hello intel my old friend

[Anonymous Coward]

I had 98 on an XP2400+ (2GHz). 10 seconds from power button to fully loaded desktop, instant shutdown, no USB peripherals for me back then (remember when USB flash drives were too small for a movie?).
I still miss it, it still had native DOS games and direct hardware access (e.g. contraptions on LPT port driven by a DOS program). It could run Doom 3 and Cygwin if you wanted.

Re:Hello intel my old friend

[Anonymous Coward]

I forgot : Windows 98 is limited to 64K in GDI resources like Windows 3.x before it. I ran into trouble with it when devs stopped caring about it! The programs glitched weirdly rather than crashing. I might have gone on with Windows ME to see if that was better but I hated how it treated DOS as a third class citizen (in retrospect, I should have tried a dual boot between 98 and ME)

Re:Hello intel my old friend

[Anonymous Coward]

AMD just released the Athlon 200GE, looks awesome. Athlon 200GE with 16GB RAM and NVMe SSD should cut through things like butter while not needing a case fan. The only thing wrong : 3.2GHz, no boost or turbo over this frequency and overclocking is locked out. So you do leave a chunk of single thread performance on the table.
Intel has the Celeron G4900 (Coffee Lake) locked at 3.1GHz, too. It's worse than the AMD : no hyperthreading and worse GPU. But then again with 8 or 16 gigs and NVMe I bet you may put this into the hand of lay people (dress it up in a small and/or thin computer case) and they'd be amazed, because they only want to do simple and useful things on a PC and so : don't wait [much] on single thread performance, never run out of RAM, never wait on I/O and never throttle or get noisy due to thermals..
I bet this is why people buy a Macbook pro to watch porn pics on tumblr.

Re:Hello intel my old friend

[thegarbz]

That's not just for Intel. That's general recommendations from processor vendors.

Re:Hello intel my old friend

[thegarbz]

That's one of the reasons AMD released "Gamers Profile" for Threadripper chips where half of the cores would be disabled and the extra thermal and electrical headroom dedicated to higher core clock boosts.

Threadrippers get a nice speed boost as a result of this feature. Though benchmarks have shown it is actually a hindrance for many of the games on the market for something like the 2700X which only has 8 cores.

Re:Hello intel my old friend

[Anonymous Coward]

Obvious fanboi is obvious.

"Reports are that the 9900K draws way more than 95 watts when running overclocked "

CPU uses more power than specified if run beyond specifications... news at eleven.

"A lot of complaints about cooling problems out there."
Mostly from people trying to sell you cpu binning and delidding services.

"A lot of doubt about accuracy of benchmarks."
Unfortunately the rampart fanboyism is fuelling the existing problems with poor standardisation with mostly youtube reviewers. The same people who were bitching and moaning about PT's review, show far less transparency with their own reviews,

Re:Hello intel my old friend

[Tough Love]

It's called a "paper release". And now "ex-fans".

Re:Hello intel my old friend

[Tough Love]

Many reports that a normal water cooler can't cool the 9900K in *normal* use. Many. Rage against it if you like, but Intel's attempt to shoehorn 8 cores into a sunset process is shaping up as mere theatre. Not going to end well.

Re:Hello intel my old friend

[Anonymous Coward]

> A lot of talk about cancelling orders and going with 2700X or Threadripper instead.

That talk might be just hot air. Threadripper works just fine for core-heavy workloads, but some tasks still perform better on Intel. Game streamers who want high frame rates and use software encoding would benefit from a 9900K. (There are other options of course, like building a capture box.)

So the 9900K is rather niche. Core-heavy users would buy Intel's 18-core offering. Gaming works just fine on an 8700K and streaming can be done with NVENC.

Re:Hello intel my old friend

[hairyfeet]

Oh I agree 110% which is why I'm in no hurry to replace my FX-8320e, 8 threads are more than enough for my workloads and with a turbo of 4.1Ghz I've never felt handicapped by GHz and if I get to that point my chip easily OCs to 4.5Ghz. And what matters to me is "how does it feel? Does it do everything I want when I want?" and my FX easily records my game streams while running a browser and background tasks so I'm happy.

But dude was trying to shill for the Pentium G5600, we're talking a dual core with HT with the hyperthreading crippled by the recent OS patches to try to mitigate Intel's lack of giving a crap about security...yeah no, just no. Not only is that chip not hot to start with but the Ryzen 3 is 1.- $10 cheaper, 2.- uses boards that support the entire line from lowest to highest model CPU/APU so has plenty of future upgrade options, and 3.- Doesn't have all the security issues the Intel chips have.

There really is no compelling reason to go for a Pentium in 2018, they are beaten in every metric except for single thread performance in a VERY limited (and growing more so every day) list of programs that can only take advantage of a single thread. I wouldn't tell someone to toss one if they had one they had bought in the past (just as I still have a Q9505s Media Center I use at the shop for looking up parts and converting VHS to DVD) but buying 1 for a new build? In 2018? Yeah really don't see a point unless you are just a Team Blue fanboy.

Re:Hello intel my old friend

[Tough Love]

some tasks still perform better on Intel

Single threaded only... about 15% better for Intel's high end parts. Not worth the nearly 100% extra cost. Ryzen 2 got nearly even with Intel's 8000 series then more than half a year later Intel puts out the aggressively clocked 9000's and gets ahead 15%. But way behind in value even for single thread, never mind multicore which is what actually matters these days with multicore rendering and video encoding now being the common loads, single core mostly legacy or things that just don't matter with the chips on both sides already so powerful. So, Intel will most probably lose the single thread bragging point in 3 months or so and will be coming from behind on Cannon Lake, which when it finally ships in volume might find itself competing against a Zen 2 increment based on TSMC's initial EUV.

See, it's about value. You need to really stretch to find a single threaded load where 15% actually impacts anybody's experience these days, but it's a no-brainer to save nearly 50% of your processor budget to spend on what actually matters, like more memory. You want buttery smooth game streaming? Get the 2700X with more than enough cores, as opposed to the imaginary 9900K that costs way too much and runs too hot even if it was actually shipping.

Yeah, a lot of talk. Pissed off talk, switching to AMD talk. You know it's true.

Re:Hello intel my old friend

[Anonymous Coward]

Shill? maybe it's what he needs.
It has a competitor, AMD Athlon 240GE but we don't know what frequency it will have yet.
Me, I've been waiting for RAM prices to crash because I don't care about the number of cores, if I want to do heavy multitasking I'm quickly limited by RAM capacity. Hence for what I do (or did before giving up) I'll take a dual core with 16GB over an eight core with 8GB, thanks.

Background CPU use depends on what software you use. Music player? that'll be maybe less than 1% CPU use. I don't need it to be written in javascript, python or have a "library" feature. Video player? When playing 480p or low bitrate 720p in a dedicated player, I'm just saving CPU versus doing it in browser. Two hundred idling tabs? That's ok. 40 text editors and 200 terminals? I won't literally do this but the CPU will bored with all these processes eating 0% CPU. Idling VM in Virtualbox wasting CPU doing nothing? That's a bigger problem but might be ok. I don't run any recent games because they are mandatory online and I don't want to broadcast an online presence this way (with an account tied to real identity).
On the other hand it's become possible to emulate an entire PC (not just dosbox) so once I have enough RAM to do web browsing *and* have RAM left for something else, I'll have a go at setting it up. You can also emulate a GameCube console, and even a Playstation 2 (never tried yet. I did try Gamecube on my older desktop which is too slow). So I do have a use case where single thread performance is more important than everything.

Encoding a video? I do that once a year or less so I don't give a crap. The other time, I encoded a clip from DVD to H264 on a Core2Duo and it took minutes (w/ highish AAC bitrate for the sound). If I needed to encode for hours, fine, you can set the CPU priority lower.

Re:Hello intel my old friend

[corydoras]

Oh, no shilling here. I was just genuinely trying to understand the disparity between the hype/price and the apparently identical single core performance -- I had to be missing something. I also hadn't looked into the cheaper Ryzen processors, and for me they seem like a good deal.

Seems like I was just living in the past where there was some relationship between price and single thread speed. Apparently now, it seems the single thread speed is pretty flat across the price ranges, and you pay more for the extra cores if you need/benefit them.

Personally, I don't, so it's a good time for me to buy fast processors for not much money.

Re:Hello intel my old friend

[hairyfeet]

Bingo we have a winnar! The difference in single thread these days between Intel and AMD is really single digits unless you are looking at the absolute top o' the line chips from both and even then you are paying nearly $700 more for a whole 12% faster single thread performance.

Nowadays with the Ryzen 3 at just $90 and the Athlon 200GE APU for just $55? Even on the low end Intel just doesn't have any real compelling choices, not when you figure in the higher cost of boards and what performance you are getting for the cost. If you want to game and don't have a GPU or your budget is super tight? The new Athlon APU can play nearly all the eSports like Rocket League and Fortnite at 1080p above 30 FPS...and that is a $55 chip my dude, and you can later pair it with something like an RX 570 or GT 1030.

Right now if someone came to me and said "I want to buy Intel, what should I get?" honestly? I'd tell them to look into the used workstation market, with so many pros dumping Xeon for Threadripper you can get some crazy good deals on Xeon 6-12 thread boxes that are cheaper than putting together a Ryzen of i5 with the same number of threads, but if all you want is a home gaming box? Ryzen 3 hands down,it has the Vega GPU, it has 4 cores with a boost of 3.7 Ghz, and if you decide you need more down the line you can just drop in a Ryzen 5 or 7 down the line with no issue.

But you were correct when all AMD had was FX, if you didn't do streaming or recording like me and just wanted a solid gaming PC? The Pentiums and i3s were good deals, its just that in 2018 the shoe is on the other foot and AMD has the better deals and the difference in games is so tiny you honestly will never notice which chip is which.

"outpaces a far more expensive Intel Core"

[Anonymous Coward]

Is all you need to know. (Oh yeah, and PCI-E lanes, and they don't have the money to bribe benchmarkers, and their PSP is a far cry from the full Intel IME. Oh yeah, and hyperthreading lol.)

Re:"outpaces a far more expensive Intel Core"

[Anonymous Coward]

Well it's more sane than Intel's crippled offerings. Intel wants you to buy the Xeon platform so you can get enough lanes to do SLI.

There is no reason why the CPU should not have sufficient lanes for two PCIe video cards, let alone four, and four NVME M2. PCIe x4 cards. So 4x16 =64, plus 16 for hard drives, and 4 lanes per thunderbolt USB-C port (most only have one right now) So you really need 84 lanes to cover all use case other than a dual processor server board.

Most people will not be doing 4 GeForce 2080 RTX's, and/or 4 NVME devices, but the fact is that many people have money to burn, and the Intel platform is not even sufficient even with the Xeon's. You can't access more than 64 PCIe lanes even on the most insanely overpriced chip.

Re:"outpaces a far more expensive Intel Core"

[Rockoon]

Someone mod this coward up.

Re:"outpaces a far more expensive Intel Core"

[Tough Love]

Far more expensive and not available. Right now Amazon doesn't even have a preorder page up, that disappeared last week when scalpers were flipping at $1000/chip.

Re:"outpaces a far more expensive Intel Core"

[Anonymous Coward]

PCIe 4 and 5 will offer some help with the lane pressure as the single lane is doubled in bandwidth once more per generation. The new trace length limits give the manufacturers nightmares already, I'm sure.

Re:"outpaces a far more expensive Intel Core"

[thegarbz]

Most people will not be doing 4 GeForce 2080 RTX's, and/or 4 NVME devices, but the fact is that many people have money to burn

Remember also we're talking about people buying threadrippers / 9900Ks. In the venn diagram of people with money to burn and customers for these products overlap greatly.

Re:"outpaces a far more expensive Intel Core"

[aliquis]

Close to 0% buy two graphics cards. Even closer for four.

As for the number of PCI-express lanes if we remove what's used to connect the chipset then the Ryzen CPU offer:
20 lanes, of which 16 goes to the GPU slot(s.)
Whereas the Intel one offer:
16 lanes.
So that's one M.2 drive connected to CPU extra on Ryzen.
However on the chipset side they offer:
X470: 8x PCI-express 2.0.
Z370: 24x PCI-express 3.0.
Typically a secondary M.2 slot on an X470 board only run at 4x PCI-express 2.0 instead, on some boards it takes lane from the graphics card and run att 4x PCI-express 3.0 with the graphics card running at 8x instead.

I wouldn't really call it much of an improvement. If you benchmark the M.2 performance especially if you use RAID the Intel solution seem to perform better.

Now technically the zeppelin die (die used in Ryzen with two CCXs) should have 32 PCI-express 3.0 lanes (of which 4 goes to chipset) so there should be 8 more available so if they could let people use those that would be an improvement.

ThreadRipper originally used two such dies for 2*32 lanes and 2*" memory controllers and Epyc used four for 4*32 and 4*2 but some of these latest ThreadRippers use four dies but don't hook up everything so they use the 2*32 lanes from two of the dies and 2*2 memory controllers from two other dies meaning the core count is the same as on Epyc but the number of PCI-express lanes and memory controllers are half and half the dies & cores lack local lanes and memory.

Re:"outpaces a far more expensive Intel Core"

[Anonymous Coward]

8x per card for SLI is rather adequate, should be as the new RTX cards have a custom interconnect through bridge connector between them (NVLink) which is kind of a short range supercomputer interconnect.
Two NVMe 4x SSD? The Intel chipset support that, both will be behind the 4x link between CPU and chipset. It's adequate. If your motherboard has another "PCIe 16x" slot or two wired to the chipset @4x, you can add more NVMe SSD, again "crippled" but really : you're sharing a 4GB/s bus for the SSDs (and shit like sound, keyboard and ethernet). This is fine, mid 2000s we still had AGP + PCI systems that shared 133MB/s minus overhead for everything save the RAM and graphics card.

Note that two vid cards is now the maximum, if you want to play vid games. When coin mining on GPU was a thing, people were doing it with 1x per GPU.
Need full bandwith for four $1200 GPUs? Sad news, you will have to buy the $400 motherboard instead of the $150 motherboard. And still have SSDs behind the chipset (still faster and cheaper than SCSI and SAS controllers with 15k rpm hard drives...)

Re:"outpaces a far more expensive Intel Core"

[Anonymous Coward]

Fact is, AMD is on a huge architectural rip and is one generation away from becoming the top dog for a long time to come. There is absolutely NO WAY I will but Intel now. AMD has my FULL support and they will have it until they have at least 50% market share. That's my philosophical goal now.... 50/50 or DIE.

AMD is back, bitches.
Now go buy AMD so you never again have to see Intel dominating the space.

Did I mention you get ECC for free with all AMD CPU's.
Fucking Intel wants you to buy Xeon or a few wimpy i3 skus for that.

Re:"outpaces a far more expensive Intel Core"

[Anonymous Coward]

"There is no reason why the CPU should not have sufficient lanes for two PCIe video cards, let alone four"

SLI now maxes at 2 and Cross-fire at 3.

Doing compute tasks rarely requires anywhere near x16 bandwidth,

What do you need 4x16 PCIe for video cards for exactly?

Re:"outpaces a far more expensive Intel Core"

[samwichse]

But still, 64 pcie lanes for the threadripper family vs 16 for the 9900k is just stupid.

We just set up a 32 core threadripper with 64gb ecc memory (yes, ecc!) and SLIed video cards and a beefy 8x PCIe SAS controller for bulk storage and a 4x nvme boot SSD. Used for scientific computing. Cost about $6k It compares pretty favorably with the dual-CPU 16 core Epyc setup we installed earlier this year for about $9.5k. Amazing.

Re:"outpaces a far more expensive Intel Core"

[thegarbz]

No "but still" about it. I fully agree with you. I was supporting the AC's statements that the type of customers using these chips are precisely the type of customers who would use the extra PCI-E lanes.

Best partner

[Anonymous Coward]

I assume they have a decent chip-set to bring out this CPUs best?

Re:Best partner

[Anonymous Coward]

x399

Re:Best partner

[Anonymous Coward]

Nowadays with both vendors the chipset is a glorified PCIe device :)

What I would love to see...

[SuperKendall]

Revised Mac Mini, offering an AMD chip.

Maybe even the redesigned Mac Pro...

To me it's been quite odd that Apple is so keen on AMD GPUs, while never using them for primary processors.

Re:What I would love to see...

[Anonymous Coward]

Because they're always too hot.

If AMD puts these out at 7nm, that puts them at a cooler TDP than potentially even Intel's 10nm.

Re:What I would love to see...

[Anonymous Coward]

What the hell are you talking about? Intel's chips are pushing higher temps these days.

Re:What I would love to see...

[Tough Love]

Ryzens don't run hot but you do need to pay attention to the TDP as always. Now we are seeing some fanless Ryzen designs coming into the market. Looks like the Ryzen 3 2200U would be fine in a NUC form factor, with its 15-25 Watt power envelope. A respectable 2 core/4 thread part. I haven't seen any Ryzen NUC-like offerings yet but there seem to be a lot of folks asking for them.

Re:What I would love to see...

[tlhIngan]

To me it's been quite odd that Apple is so keen on AMD GPUs, while never using them for primary processors.

Easy. Part shortages. One problem during the PowerPC era that Apple had was that neither Motorola (now Freescale) nor IBM would supply Apple with the processors Apple wanted. It was so bad that there would be delays for weeks of the top end models that Apple offered because the yields were terrible. This happened so consistently it was predictable - if you wanted a high end configuration, you reloaded the Apple Storage page and got your order in ASAP.

Freescale was not interested in fixing the shortages, which is why Apple went to IBM for the G5 processors, hoping IBM would give Apple more priority.

The reason Apple went with Intel was because Intel has sufficient capacity to actually produce the chips Apple wants. AMD unfortunately still had plenty of yield problems - evidenced over the years of how hard to get the high end chips were. Should Apple have picked AMD, AMD would've suffered trying to fulfill Apple's orders to the point where the market would've been flooded with unwanted processors that didn't make the cut.

ATI though didn't have those issues. And the only reason Apple went ATI was because nVidia's CEO blabbed about being Apple's supplier a couple of days before a keynote. Surprise surprise, said keynote now featured computers with ATI GPUs.

Clearly AMD has learned their lesson with Ryzen as I haven't heard of part shortages at any level - perhaps the old ATI team taught AMD how to design parts with higher yields.

The GPU integration has me scratching my head. AMD's integrated slaughters Intels in any fair (equal $) comparison. I dont see how the deal with Intel benefits AMD.

Intel is the #1 supplier of GPUs in the world. But as we all know, they suck. Even the higher end Iris models are OK, but generally not as performant. This deal combines an Intel CPU with an AMD GPU giving a single package integrated solution, something Apple seems to have their hand in making it possible.

It benefits AMD in that the #3 GPU supplier in the world is partnering up with the #1 CPU maker in the world. With this, AMD has an advantage that nVidia does not have - namely being able to provide Intel (#1 GPU supplier) with a very interesting competitive part. nVidia is stuck, because AMD provides integrated GPUs for both chipmakers, themselves and Intel. And Intel is happy because they now have a third option for highly integrated systems - you could go with a low end Intel GPU, a low but higher performance Iris Pro solution, or you can get a nice decked out CPU with AMD Vega graphics. It might not blow the pants off an nVidia GTX 2080 Ti, but for a mobile system, it offers compelling value.

And Apple is behind it - they want higher end (better than Iris Pro) graphics in a single package solution for stuff like their MacBooks and such which are space constrained.

Re:What I would love to see...

[Bert64]

Intel have been offering integrated third party GPUs for a while, several of the atom chips were available with a PowerVR GPU.

Re:What I would love to see...

[Anonymous Coward]

The Ryzen 2200U, 2500U, 2700U are fairly weak on CPU performance. Driving CPU + GPU at 15 watts is a bit too much for them, with the powerful GPU present.
They'll do better at 7nm (their successor, more accurately) or failing that the performance is good at 35 watts.
That or they should cripple the GPU at very low speeds to give a couple watts more to the CPUs...
I am pretty sure the Mac Mini will use Intel CPU with integrated GPU at 15 watts, a minor CPU respin called Whiskey Lake. These are 2C/4T and 4C/8T parts with some crazy high CPU boost speed (4.x GHz, 3.9GHz for the 2C/4T)

Re:What I would love to see...

[Anonymous Coward]

That's true :). It was in the time when Intel wanted to go into smartphones. And they did, the Asus Zenfone used it and it was good but otherwise not remarkable. Smartphone is a rat race and some of the vendors do yearly release of low end, mid range, higher end, flagship. Intel like nvidia felt they were more at home with higher margin chips. I've just remembered that nvidia had a smartphone chip, the Tegra 4i. I bet this is pretty forgotten now.

To contrast, AMD GPU + Intel CPU mobile package is very much a high margin affair (though expensive to build as well. Everything with HBM memory is expensive). It's mainly seen in a luxury Intel NUC.

Re:What I would love to see...

[Tough Love]

The Ryzen 2200U, 2500U, 2700U are fairly weak on CPU performance.

Compared to what, have you got numbers? Looks like 2200U clobbers Celeron N3050 in performance while being in the same ballpark in power consumption. Didn't look deep, but this does suggest it has a place in SFF.

Re:What I would love to see...

[Anonymous Coward]

Compared to Intel Kaby Lake and Coffee Lake? These are what the Ryzen "U" processors compete against, in laptops. But maybe I remember wrong.. Very wrong maybe. Its gaming performance is crippled if the laptop cooling is not enough.

Desktop motherboards with a soldered laptop CPU are a rarity at least on the AMD side, all I can think of is the Asrock QC5000 which had a soldered Kabini (different from the Kabini on socket AM1 though functionally the same, one was made at TSMC and one at Global Foundries)
If you can afford 35 watts then the CPUs can be bought at least, every AMD ending in "GE".
Should be ok to go with a small ITX case (as opposed to a "gamer ITX" case) esp. if you're not using a 2.5" HDD.

ATI tie-in runs deep

[evanh]

While I agree that Apple giving AMD a spin would be welcome, I can also see the reasons why there hasn't been any movement:

  - Apple working with ATI is positively ancient. AMD buying ATI didn't affect that. So that explains the continued use of the AMD Radeon line of GPUs.

  - The transition from IBM Power to Intel Core CPUs was done in classic Steve Jobs flamboyance. And with that Apple got preferential Intel treatment that had been the domain of Dell. (Whole other story there me thinks.) Apple won't be in any rush to jeopardise this position.

  - You'll note also the deal between Intel and AMD for GPU integration. I would expect this deal is at the behest of Apple.

Re:ATI tie-in runs deep

[Rockoon]

The GPU integration has me scratching my head. AMD's integrated slaughters Intels in any fair (equal $) comparison. I dont see how the deal with Intel benefits AMD.

Re:ATI tie-in runs deep

[Voyager529]

And with that Apple got preferential Intel treatment that had been the domain of Dell. (Whole other story there me thinks.) Apple won't be in any rush to jeopardise this position.

Just spitballin' here, but something tells me that whatever preferential treatment Apple got when they switched the Macbooks over may well be back in Dell's hands.

Yes, the Macbook is still the darling laptop of college kids...but Apple's bread and butter has been iOS devices, which have all been in-house CPUs for years (and ARM before that). Intel doesn't get a slice of that very, very large pie. Moreover, Apple versions of Intel processors are at least somewhat custom runs, since they are all soldered in now and don't take sockets. One could argue that most laptops are designed this way and have been for some time, but their desktops (save the old tower style Mac Pro) all have soldered chips too, so there's less volume to be had there.

By contrast, Dell laptops still sell pretty well in corporate environments. Their other major division is servers, virtually all of which house Xeon processors. Dell may not be selling iPad quantities of PowerEdge units, but they're still selling a respectable number with 1/2/4/8 of Intel's CPUs.

If Apple has any preferential state with Intel left, it likely isn't due to their revenue...and when your best selling products use somebody else's silicon, that's not the fast track to endearment.

Re:ATI tie-in runs deep

[RatherBeAnonymous]

And with that Apple got preferential Intel treatment that had been the domain of Dell. (Whole other story there me thinks.) Apple won't be in any rush to jeopardise this position.

I think you are correct considering that A) Apple sells new machines with 4 to 5 year old CPU models, and B) Apple is supposedly designing their own desktop and laptop chips to be rolled out in about 2 years.

Per core licensing

[jezwel]

So long as these hefty core counts per socket don't end up in my Per Core licensed devices, I should be OK....

Re:Per core licensing

[torkus]

And I think that's what AMD is missing here. Cramming 24 or more cores into a CPU has already long passed diminishing returns for anything but very highly optimized parallel-thread applications. ... and those same applications typically cost FAR more per core to license than the CPUs themselves. For the average consumer and even the performance kiddies I don't see how they're really winning much in real-world terms.

Re:Per core licensing

[Anonymous Coward]

The world is a great deal bigger than you think.

There are _plenty_ of parallel workloads out there than Oracle or what you're thinking of. I have a hard time understanding why you're wanting to deny "average consumers" playing with Blender or encoding video or why not some of the FEA packages out there, just to give a few examples.

Not that the "average consumer" does much more than browse and watch YouTube or Netflix, for which you have absolutely no use for a Threadripper or one of Intel's counterparts.

And who cares about "performance kiddies"? Those are both few and completely clueless anyway. They will go for Intel either way no matter what because that's what everyone else uses. They neither have the knowledge to sort out what each chip is actually good for or not, what's a doctored benchmark, what's a realistic workload, nor even remotely understand what a compiler is, what it does, how they differ and how they can, and are, manipulated.

Threadripper is for serious workloads, not for playing games. "Highly optimized parallel-thread applications" is _exactly_ what it's designed and marketed for. If doesn't match your use-case, that's what Ryzen is for, or EPYC in case Threadripper isn't serious enough. AMD isn't missing anything. It's you who are thinking like a gamer kid.

Re:Per core licensing

[Anonymous Coward]

OTOH that is a bit of hen-egg situation.
Why would anyone bother with making the code parallel threaded if the user base doesn't have enough cores to make it worthwhile.

With the abundance of cores it makes more sense to spend some effort in parallelizing.
Maybe some day we will be able to stream video while browsing a javascript heavy page at the same time.

Re: Per core licensing

[LordWabbit2]

Erm, I do that now, and we also code for multiple cores / threads. I bought my I7 ages ago, second or third generation I think, and for home use I have to admit, it idles really fast. For work, well I've made my I7 sweat, but it took some effort.

Re:Per core licensing

[Anonymous Coward]

Now there's a subtlety : 32-core Threadripper sucks ball under many benchmarks in Windows, and we can assume the same goes for the 24-core model.
The reason is Windows has poor NUMA support and this product is like a NUMA-NUMA. It's a CPU made of CPUs, some of which have no memory directly connected to them so they have to do a memory hop for anything.

I don't know if Windows 10 will improve and support this properly, otherwise you really have to do proper workload-shopping if going with a 24-core Threadripper for Windows, or forget about it and get the 2950X.

Re:Per core licensing

[Anonymous Coward]

I'd say anyone investing that amount of money in a CPU either has an extremely clear use case where it's a perfect fit, or they are fools. And Windows isn't the be all end all, increasingly so between forced reboots and broken updates. Having your computer rebooting 2/3's through a multi day job because it thinks updates are the most important thing ever is always a ball.

Re:Per core licensing

[Anonymous Coward]

I had Adobe users in mind especially. They don't typically wonder about how their L3 caches are split or what is asymmetric NUMA. They probably care more about what they will wear for the wedding they're covering, or about finding corporate clients.
It's a CPU at "only" $1299 that promises to be like 3x faster than some Apple machine and it is except when the results are fucked up.

Laptops are a big thing, though. I know a guy who does graphics works and saw he has a thick gaming laptop (the kind with "warriorz" esthetics). I thought, good call! It would be a bitch to take in bed, but if it sits on a flat surface then it's like any laptop, except the heavy duty cooling alone is sure to make it more powerful.

Windows must die?! Yes. But like it or not the "workstation" role is filled by Apple and Windows. Maybe we could have had BeOS or Solaris on desktop x86 running all this kind of stuff but we didn't.

Selling More

[JBMcB]

The GPU integration has me scratching my head. AMD's integrated slaughters Intels in any fair (equal $) comparison. I dont see how the deal with Intel benefits AMD.

Because Intel still sells a lot more desktop chips than AMD. AMD probably won't be making any serious moves in the low to mid-range desktop CPU market any time soon. Intel has that locked up. So you may as well make money selling AMD graphics on those low end chips.

There is something to be said for making strategic decisions on not partnering with potential competition. There is also something to be said for selling as much product as you can to make money.

Re:Selling More

[Tough Love]

AMD probably won't be making any serious moves in the low to mid-range desktop CPU market any time soon. Intel has that locked up.

What makes you think that? For example, HP Pavilion with Ryzen 2400G, $488. Looks like low to midrange to me.

Re:Selling More

[Anonymous Coward]

There is something to be said for making strategic decisions on not partnering with potential competition.

Competition is something that doesn't happen.

There are essentially 2 vendors to pick from and they are making deals.
You need something like 20+ for the market to start working as intended.

Re:Selling More

[hairyfeet]

Dude have you SERIOUSLY not been on YouTube lately? Every single reviewer is singing the praises of the new Athlons for the low end, those things can play most of the eSports titles and a surprising number of the mainstream AAA titles and the thing can be had for $55 so you can bet the desktop makers are gonna be sucking those puppies down like Tic Tacs, especially now that Intel is having supply issues.

I think Intel is seriously gonna be hurting in 2019 which is why we are seeing the same old shenanigans like we saw during the P4 with the rigged benchmarks and crap,at this point the really either have to just murder their server sales (and lose serious profits) by moving more and more Xeon chips into the i series or continue to lose sales as AMD is frankly just offering better deals.

The uptick for the end user though is I'm seeing Xeon workstations just get dumped on the used market for some crazy cheap prices as the pros are all switching to Threadripper so if you want a cheap gaming PC? Those Xeon workstations are quite nice, not gonna be doing Intel any favors though.

Lanes

[JBMcB]

Are 16 lanes per GPU that important? There was an old benchmark from a few years ago that compared SLI using 2x16 to 1x16/1x8, and there was hardly any difference. Granted this was on relatively old hardware (I think 8xx series GeForce) but from what I remember the important part was bandwidth loading textures into the GPU. Since modern GPUs have gobs of RAM, and the faster SLI bridges let them pool memory better, it's less of an issue.

Re:Lanes

[Anonymous Coward]

That depends on what you're doing. High resolution, multi screen etc, will quickly swamp the buss. Add to that networking, a couple of NVMe units and a few USB devices connected via pcie and you're quickly running out of pcie lanes.

Re:Lanes

[aliquis]

There never was any desktop 8xx parts, only laptop, desktop went straight from 7xx to 9xx.

The PCI-express version is still 3.0, may have been 2.0 when Linus Tech Tips or whomever tested it back then but the graphics cards has become faster.

On the other hand AMD has abandoned CrossFire branding completely and SLI is pretty dead/unused too. In DX12 you may be able to even force rendering onto two cards but really close to no-one is using two cards for gaming and very close to no-one is using four for that purpose.

SLI didn't worked with a bunch of games and even when it work people complain on micro stutter so personally I kinda feel it may be more worth to get the most expensive model and considering those sit at ~$1300 now you gotta ask yourself how many spend that and how many are willing to spend two-four times more than that? Also those cards can do 60+ fps 4K gaming alone.

For textures the compression has improved which Nvidia argued compensated for the lower memory bandwidth. AFAIK SLI doesn't let you share RAM between the cards, I don't know what the SLI bridge actually do. The new cards use NVLink and as such at-least for the professional cards you can use all the RAM added up together and on the professional cards the connection is very fast but on the consumer cards it's slower but I don't know if they still add the VRAM together and if the typical scenario is just two cards maybe the lower bandwidth doesn't matter all that much.

Not SLI:
https://www.techpowerup.com/re...
8x vs 16x:
AC:O 1080p: 104.7 vs 114.5 fps.
BF1 UHD: 95.8 vs 97.6 fps.
CoD:WWII 1080p: 207.7 vs 217.0 fps.
Civ VI UHD: 115.3 vs 122.7 fps.
And so on.
They lost 2-3% using 8x all in all over the various resolutions, that was just for a single card but hardly worth being upset about.

SLI:
https://www.gamersnexus.net/gu...
Ashes of the Singularity, UHD Crazy:
16+16: 127.2 / 47.7 / 42.8 (avg, 1% low, 0.1% low)
16+8: 125.1 / 46.9 / 41.8 (but in what situation will you ever have 16+8?)
8+8: 106.9 / 40.9 / 37.3
So with NvLINK/SLI it matter more than for a single card.

As for scaling:
NvLink 2080Ti Sniper Elite UHD High DX 12: 209.8 / 145.1 / 132.1
2080Ti FE: 108.2 / 93.2 / 91.4
So a very good result there.
SLI 1080Ti: 170.3 / 104.8 / 97.4
1080Ti: 86.6 / 75.1 / 72.2
So very good results both with new cards with NvLink and the old cards with SLI in that DirectX 12 title.

I don't really know what I wrote when I started this post but for a single card 8x seem to only make 2-3% difference, for two it's more like 20% difference and in a game which has great DX12 and scaling results using two cards whatever SLI or NvLink worked great.

Re:Lanes

[Anonymous Coward]

You get 16+8 on high desktop Intel platform, when using the cheaper processors that are crippled to 28 lanes. (they're launching i7 9000 X series that won't have this limitation anymore)
e.g. i7 5820K was an excellent CPU that gave you six cores and twelve threads quite some years ago (somewhat cheap except for the motherboard), but 28 lanes vs 40 lanes on i7 5930K.

Regarding NVLink iirc there's a single NVLink between two RTX 2080, and two NVLink between two 2080 Ti (doubled bandwith). But anyway it is a good idea to go single card.

Re:Lanes

[aliquis]

Yeah, didn't counted anything just thought it was a weird number.

I was looking a lot into the i7 5820K back in late 2014 wanting to buy one even though people said 4790K was better and 4690K enough. I then regreggeted not buying one because I was stuck with my old junk because the SEK tanked vs the soaring USD and deals went away.
The processors cost basically the same but the X99 and DDR4 cost twice as much as Z97 and DDR3.

By now latest rumors are Epyc will have 8 core dies of 8 cores each and 32 MB cache and then a central non-core die for other things. The octa channel and 128 lanes may still be a thing I don't know but that's what Intel got to compete with ..

Going up in cache from 16 to 32 MB / core die use up more space then again they did 8 core dies onto 14 nm before and 7 nm now so they had room to spare. Yields of such a 64 core + everything else 9 dies behemoth must be much much better than if everything was on one die.

Oft overlooked

[NewtonsLaw]

About 6 months ago I built a new budget video editing rig. I was torn between going with an i7 8700 or a an AMD2700 but opted for Intel because of QSV.

QSV allows for decoding and encoding H264 and H265 video in hardware using the on-chip video hardware. It's brillant watching my 6-cores idling while rendering 4K video into H265 files at realtime speeds. Try that with your AMD processor :-)

However, these days I'd probably go for the 1950 Threadripper (cheaper and almost as good as the 2950 because those extra cores *are* useful in good video NLEs such as Davinci Resolve.

Re:Oft overlooked

[Tough Love]

It's brillant watching my 6-cores idling while rendering 4K video into H265 files at realtime speeds. Try that with your AMD processor

Doesn't QSV rely on the GPU? In which case the Ryzen 2400G's Vega 11 considerably outpowers the 8700k's UHD 630.

Re:Oft overlooked

[thegarbz]

I am very curious as to why QSV exists on desktop processors. QSV makes sense for mobile chips and AMD's APUs have a similar functionality. But frankly QSV is a slow dog compared to offloading rendering onto the compressor in even an old GTX1060.

These features exist in video hardware, so it makes no sense to duplicate them in the CPU, espeically given the video cards typically have a shorter lifetime (for the power hungry) than a typical CPU. (At least for me personally, I've bought twice as many graphics cards in my time than CPUs and no I don't run SLI).

So in summary:

Try that with your AMD processor :-)

No thanks. I'll do it on my NVIDIA GPU. Also realtime speeds? You're ONLY achieving realtime speeds? Shouldn't you be getting close to double realtime speeds at 4K (assuming 60fps progressive content) from a dedicated video encoder?

Re:Oft overlooked

[thegarbz]

No. Yes. No. Wait!

QSV is in part of the CPUs which have a dedicated video engine. AMD's APUs have the same thing (called VCE). However this feature is quite irrelevant on a high performance desktop which will have a dedicated GPU in it anyway.

Re:Oft overlooked

[ytene]

Is it to cater for system-on-a-chip systems, or maybe laptops, where there is no dedicated GPU? For example, something like we see in Intel's Core i7 7700T and 8700T processors, which bundled Iris graphics with the core processor on one chip?

Re:Oft overlooked

[Tough Love]

Right, really only relevant to laptops. But I bet there are way more people encoding video on laptops than desktops.

Re:Oft overlooked

[thegarbz]

Yes exactly. As I said this is why AMD APUs have the same functionality, as does my several year old Core i5 with Iris graphics. However it makes no sense on chips in the class which we are discussing.

Re:Oft overlooked

[DigiShaman]

They exist on desktop processors because many business use the iGPU/APU only, yet streaming video benefits being able to decode H.264/H.265 in hardware. For the DIY gamer or Workstation build, yes, it doesn't make sense as you're going to have a GPU card installed. But via volume of new PCs sold, yes, absolutely there's a market beyond your myopic perception of the industry!

Re:Oft overlooked

[thegarbz]

You missed my point. What kind of a weird arse computer build specs out something like a Ryzen Threadripper and then doesn't have a dedicated GPU. If you think it's the top of the line CPUs that are finding their way to industry then it's not my perception of reality that needs to be questioned.

Posted from my work PC with a low end processor where it makes perfect sense to include VCE on the AMD APU

Re:Oft overlooked

[Anonymous Coward]

I did that with my myth server while running 4k 10bit HDR with an inexpensive nvidia gtx960 that handled it all in hardware. Why would you want a CPU to do what the GPU is DESIGNED for ?

Re:Oft overlooked

[Anonymous Coward]

From wikipedia:

Intel® Quick Sync Video is Intel's brand for its dedicated video encoding and decoding hardware core. Quick Sync was introduced with the Sandy Bridge CPU microarchitecture on 9 January 2011, and has been found on the die of Intel products ever since.

The name "Quick Sync" refers to the use case of quickly transcoding ("converting") a video from, for example, a DVD or Blu-ray Disc to a format appropriate to, for example, a smartphone.

Unlike video encoding on a CPU or a general-purpose GPU, Quick Sync is a dedicated hardware core on the processor die. This allows for a much more power efficient video processing

---

TLDR looks like its intel only and its not a cpu or General purpose GPU but an on die bit of specialised hardware!

Re: Oft overlooked

[nehumanuscrede]

" You missed my point. What kind of a weird arse computer build specs out something like a Ryzen Threadripper and then doesn't have a dedicated GPU. "

CPU based render farms.

Re:Oft overlooked

[Anonymous Coward]

When Intel Skylake came out I read the in-depth reviews, and for this the significant part was that encoding on the Quicksync video engine was done with a much higher quality than on previous Quicksync (Haswell/Broadwell and earlier).
Encoding in software still is better (x264 or x265 gets you there) but there's no obvious washed look on Skylake's Quicksync.
Encoding on GPUs, like older Quicksync takes shortcuts. I.e. you can gloss over encoding features and still get a perfectly legal h264 file, so that you get high encoding performance (in terms of megapixels or gigapixels per second) at the cost of quality on a lower end hardware encoder or when encoding on GPU.

To make things more complicated there are two ways to encode on GPU : with the dedicated hardware encoder, or by using the GPU shader units as a processor (this is CUDA, OpenCL etc.). Dedicated hardware encoders on GPU are very powerful but have low latency in mind i.e. game streaming and "cloud gaming" or supporting professional software on thin clients. (We don't care about latency for background task encoding). Using the shader units means avoiding excessive branching in code or otherwise keeping the code friendly to GPUs, so you do leave quality on the table otherwise the GPU software encoder would run like crap.

Re: Oft overlooked

[thegarbz]

CPU based render farms.

Render farm ... where hardware h.265 encoder is a critical application... A friend of mine owns a landscaping business and could rent you a nice backhoe, not that I think you're not doing a fine job digging yourself and your argument in a hole, but still I'm sure he'd offer his services.

Re:Oft overlooked

[philmarcracken]

I'm glad you find encoding to licensed codecs so enthralling.

Re: Oft overlooked

[Anonymous Coward]

HW based encoders are nothing more than ASICs. They're exceedingly fast and efficient per watt. But, they're ridged. Software based encoders offer more flexibility in tuning to get the best quality encoding per unit of storage, but it's slower by orders of magnitude.

QSV and NVENC are perfect for real-time encoding. But for watching a movie, you only need to decode. For that market, quality is preferred and it only needs to be encode once anyways.

Re:Oft overlooked

[Anonymous Coward]

In graphics. Does it overpower it in video encoding? Does it overpower it in software support for said video encoding?

Re:Oft overlooked

[Tough Love]

In graphics. Does it overpower it in video encoding? Does it overpower it in software support for said video encoding?

Don't know. Maybe. The Ryzen APUs have VCE 4.0, also a video ASIC, as part of the Vega GPU. I don't know how the GPU figures into it, if at all, but one thing seems clear: QSV is not a reason to stick with Intel, even if transcoding is your main thing.

Re:Oft overlooked

[Anonymous Coward]

Discrete GPUs have hardware encoders (GTX series from NVIDIA, some others from AMD). That's how I'm able to convert video on my old i3 at a reasonable rate.

Most people building a 2700 system are probably going to have a gaming-quality GPU as well, so it shouldn't have been an issue.

That said, it would be nice if AMD could put out a six-core processor with a discrete GPU. That's the ideal point right now for generic computing. iGPUs are perfect for office work and allow gamers to wait out increases in GPU prices.

Re:Oft overlooked

[thegarbz]

I'm glad you find encoding to licensed codecs so enthralling.

I think you'll find he's not enthralled, but rather like most people simply doesn't give a shit.

Re:Oft overlooked

[Anonymous Coward]

integrated GPU, you mean?
I think with 7nm Zen2 in 2019, AMD can easily do an APU with eight CPU cores. Can't make the GPU very much big anyway, because it will be starved by the low memory bandwidth. But I'm not sure they'll do.

Intel C3958

[Anonymous Coward]

2920X = $649 i9-9900K = $580

Frankly you'd be nuts to go for the intel chip when you can get 50% more cores AND an upgrade path to 64 cores for around the same price.

Intel Atom® Processor C3958, 16-core, 16-thread (no hyper-threading) low power (full load 47W), fan-less design, with 16MB cache, up to 2.0GHz, can access up to 256GB of RAM, and can be used as a edge-computing device or a dedicated web server.

https://ark.intel.com/products...

Server motherboards from Supermicro, TYAN, or Gigabyte, with CPU attached, for less than $800.

https://b2b.gigabyte.com/Serve...
https://www.tyan.com/Motherboa...
https://www.supermicro.com/pro...

When can AMD offer us something similar?

Re:Intel C3958

[Anonymous Coward]

Today.

Actually not the fairest of comparisons, but I'm not going to do the parent poster's fucking job for them.

Re: Intel C3958

[Anonymous Coward]

Today.
Get an 8 core hyperthreaded ryzen like the 1700 or so, for 250 usd. Then underclock it from3+ ghz to 2, so instead of 65W it uses less than 40.
Pair up with a 250 usd mobo, and you're good to go, with 300usd left for fun stuff.

Re:Intel C3958

[Anonymous Coward]

The Atom C2000 series were a failure. Some design or fabrication flaw made them die after a couple year. No reason the C3000 series would be the same but what a surprising disaster it was. Intel could keep this low key because I don't think such "microservers" have much if any success.
One little theory I read about is : you need to keep your application response time under 100ms or 300ms or X ms - say, 200ms might be nice as a maximum because it's about human reaction time. Your developers are mediocre (most are) and developing some business web crap in Java, .NET, PHP or Node.js. Halving your single thread performance will just double your reaction time and make you look bad, all for saving $1000 in server hardware and a few watts. The code is atrocious, but a twice faster CPU is much cheaper than recruiting and paying good devs. How many servers run the default super fast Xeons with the same cores as on laptop and desktop, 80%?

Regarding the question you asked :) AMD does have a product they call Embedded Epyc. Nah, Epyc Embedded. Here it is https://www.amd.com/en/products/specifications/embedded/8161

Looks like a soldered Ryzen but with reconfigured I/O (trades PCIe lanes for 10GbE). I suppose it could support registered memory but it isn't stated (note that with new unbuffered memory 128GB is possible, even on a low end desktop). It's more akin to Xeon-D than Atom C but that might be fine, maybe Xeon-D was more popular.. Epyc Embedded 3000 might have an edge there with higher single-thread performance.

The product existing isn't enough, it might sell if/once AMD establishes its server presence. ARM servers also exist and despite being a bit hyped, they just haven't been successful yet (at worst, same fate awaits the Intel C3958). Small slow manycores is a niche and is just what you want in that niche, but still a niche.

Re: Intel C3958

[Anonymous Coward]

The 1700 is good and the 2700 has even better performance per watt (relatively crippled on absolute performance, because it really takes its 65W TDP seriously).
It would be perfect if it had a server motherboard to go with it with basic built-in VGA (there's none in the CPU) and guaranteed for real ECC (it's ambiguous).

Production runs are being binned. Faulty 32 core u

[Anonymous Coward]

Intel does the same. Smart practice. I'll probably get a 24 core myself.

Large core count has limited value

[jd]

What matters is cache size, L2 and L1. Losing a few cores and bolstering cache will improve performance in quite a lot of cases.

The Key is that threads don't talk that much. The amount of shared information needed to justify cores in close proximity and a huge shared cache isn't there a lot of the time.

Cheaper SMP - not difficult with PCI-E's design - would leave much more room for the critical L1 cache, reduce the heat burden on a CPU, and potentially quadruple the number of cores (since 4-way SMP is not too bad).

Close proximity on silicon only matters when you're communicating between units. Totally independent computation can be done anywhere. Port Linux to SystemC and compile it to an ASIC if you want. Should run fine, even if taking no cores at all.

There's zero relationship between half the system services on Windows/Linux and the applications being run, so there's no gain through physical proximity. There's no latency issue to resolve.

On the other hand, those same services reduce the L2 cache space your applications have available, so your applications are fetching from main memory more than they have to, just so that the services can poll for any work, scan your computer for viruses, etc. Not essential activity when playing Elite:Dangerous or Kerbal Space Program.

Not dissimilar to splitting off cores that were connected with graphics, now known as GPUs.

I'm thinking back to a model of an Amdahl mainframe. Similar sort of concept, segregated processing.

Re:Large core count has limited value

[DigiShaman]

There are many problems that can be solved multi-threaded yet share the same memory pool. NUMA is a bitch when you're having one CPU cross over to another bank of memory located on the other side. It's why when possible you want to avoid NUMA spanning. AMD's application attempts to prevent that.

Re:Large core count has limited value

[PhrostyMcByte]

What matters is cache size, L2 and L1. Losing a few cores and bolstering cache will improve performance in quite a lot of cases.

Do you have any links researching this? Outside of CPUs cache very quickly has diminishing returns, and I'd be curious to know what effects it has on CPUs.

I think one of the best things Threadripper has done is to finally introduce NUMA to consumer processors. It's going to take a hit because a lot of code still hasn't updated to support it, but it's a smart long-term add.

Re:Large core count has limited value

[Anonymous Coward]

I'm not sure I understand where you want to get at, but AMD has their coherent link "Infinity Fabric" and some of the I/O lanes from the Zen die can be configured as PCIe or Infinity Fabric. Such that a single Epyc CPU supports 128 PCIe lanes, and two Epyc CPU also support 128 PCIe lanes. Half the lanes went into the interconnect between both sockets.

Inside a single Zen die, you have two four-core blocks already (and some performance loss if migrating threads between them or reading the L3 cache from the other block).
Also a Threadripper or Epyc CPU has several Zen dies.
So, while I don't understand what you wish for, AMD has a kind of "Russian dolls" model there and it seems to be working.
We've not seen yet a GPU or GPUs plugged into that fabric, save for the current APU (2200G, 2400G, 2500U, 200GE.. all a single design, the Raven Ridge)

Re:Large core count has limited value

[Agripa]

What matters is cache size, L2 and L1. Losing a few cores and bolstering cache will improve performance in quite a lot of cases.

This is easier said that done, especially for the L1 cache. L1 cache size is limited by the cycle time and load-to-use latency of the instruction pipeline. If they could make the L1 cache larger without sacrificing clock rate and load-to-use latency, they would.

This is why complex out-of-order instruction pipelines *must* be used to achieve high clock rates. They increase the allowable load-to-use latency which increases the allowable latency of the L1 cache. But good luck pushing the load-to-use latency above the current 4 cycles without just burning lots of power for no gain.

Higher layers of cache are not tightly coupled to the instruction pipeline but the same considerations still apply.

Re:I saw "thread ripper"

[Anonymous Coward]

Do you really fantasize about creimer's pants that often? If so, maybe you should tell him how you feel.

Is There A Pattern In The Data

[ytene]

On the one hand, it is always good to see innovation and improvement in technology. Kudos to AMD and Intel for continuing to develop and evolve new technologies.

On the other hand, am I the only one that thinks that both companies have completely lost the plot when it comes to model/variant naming conventions?

In fairness, a big part of the problem is not entirely the fault of the chip makers... As the core computing world (desktop/mobile/server) matures, we are seeing the most successful companies achieve dominance through an ability to tweak their designs to more closely match the demands of their clients. Everything is up for optimisation - clock speed, core and thread counts, L1 and L2 cache, TDP, power consumption, the works. This generates a *lot* of different processor models.

The problem is that when many of these chip permutations then make their way in to the retail channel, the resultant model naming conventions and "chip families" just result in endless confusion. Whilst it's also fair to say that it is not too difficult to figure out low, medium and high performance models [start by looking at prices within a given range, then dig for details], we're increasingly needing to become chip specialists who have a very clear idea of our intended use cases if we want to have confidence that we've bought the best chip for our desired task profile.

I'm curious to know if slashdot readers think this is a fair criticism and/or whether there would be any interest in having a more uniform way of assessing the relative merits of different chips. For example, if I compare the Intel Core i7-7700T with the Core i7-8700T, not only is the move from 7th generation to 8th generation relatively easy to spot, but when we look at the specifications, then with pretty much everything except the base processor frequency, we can see the improvements delivered by the later generation. That sort of direct comparison just doesn't seem possible with the latest product announcements...

What would you do differently? Or are the current naming conventions from AMD and Intel easy enough to follow?

Re:Is There A Pattern In The Data

[thegarbz]

Naming conventions stopped being relevant beyond the target for the machine a long time ago, and by that I mean designations for overclocking or for laptops etc.

In general there are so many different variants and feature sets in any CPU suited to such a large number of different ideal workloads that no sane naming convention could keep up. Not unless you start following industrial product naming conventions such as:

AMD-GENERATION-CORECOUNT-PACKAGE-SMT-FEATURE1-FEATURE2-FEATURE3-FEATURE4 . etc .

Throw another set of cores on the die

[Anonymous Coward]

The obsession these days is more cores I guess. I'm sure its somewhat useful for a small percent of PC users. But its sort of like having a 190 MPH car to drive on a 70 mph road. Its there if you need it, but will you ever use it?

Re:Throw another set of cores on the die

[Bert64]

There are many advantages to a car capable of much higher speeds than the legal speed limits...

Operating an engine close to its maximum power output is inefficient and increases wear and tear, if a car is capable of 190mph but it spends most of its life at 70mph then there is very little stress on the engine and it's likely to last a long time.

A car with a higher top speed typically has better acceleration too, you may not drive any faster than 70mph but your time to go from standing to 70mph will be lower than a slower vehicle. Sometimes the extra performance can also be useful in an emergency too.

Plus there are places where you can legally drive at higher speeds (racetracks, german autobahns etc).

Re:Throw another set of cores on the die

[Anonymous Coward]

The obsession these days is more cores I guess. I'm sure its somewhat useful for a small percent of PC users. But its sort of like having a 190 MPH car to drive on a 70 mph road. Its there if you need it, but will you ever use it?

A lot more than you'd think, actually.

I have an older version of the AMD CPU, it presents as an 8-core, but it's really a hyperthreaded 4-core.

But, you can rip a CD, play music, surf the web, and do several other things at the same time without having the CPU become busy. It leaves capacity for the OS to do whatever it's doing, and for the user to do what they're doing.

The ability to have better concurrency helps out in a lot of ways, and I have rarely actually saturated the CPU.

On the other end of the scale, I also run at least one VM at home as a sandbox, and at the office I frequently do things in Excel which peg all 4-cores in my work laptop at 100% for a few minutes at a time. For things like databases and highly concurrent applications like web servers, the more cores you have the better.

Most people will never do things which make that many cores completely busy, but it essentially lets you have better multi-tasking, and the things which can use it can have loads of CPU sitting around.

My next machine I expect to have one of these CPUs in it, because for me the extra cores really come in handy.

I'm not mostly doing super CPU intensive stuff, but I do run a crapload of stuff at the same time. And for that, more cores/threads means a more responsive machine.

You'd probably find if you did the same things on a single core machine versus having lots of cores that overall the machine is much more responsive.

Re:Throw another set of cores on the die

[drinkypoo]

Operating an engine close to its maximum power output is inefficient and increases wear and tear, if a car is capable of 190mph but it spends most of its life at 70mph then there is very little stress on the engine and it's likely to last a long time.

They solve this problem now with either 7-speed (or more) transmissions, or with CVTs. You can't compare computers to cars. It never works. Stop it.

A car with a higher top speed typically has better acceleration too,

They're solving this problem with mild hybridization. The starter and alternator are replaced with one belt-driven motor/generator which gets the vehicle moving while the engine is stopped, eliminating the drawbacks of auto start-stop. It also can torque fill.

Most people don't need lots of CPU, by modern definitions. The average user is just web browsing. They can get away with about four decent cores. What they really need is a boatload of RAM. More RAM is always better, unless you're trying to hibernate.

Re:Throw another set of cores on the die

[Anonymous Coward]

>Its there if you need it, but will you ever use it?
Yes, I will.

Re: Throw another set of cores on the die

[nehumanuscrede]

My next system is a 16 core liquid cooled Threadripper paired with 64GB of memory, a pair of 1080ti GPU's and a few M.2 SSD's. Should keep the system relevant for a few years at least ( which is the goal )

Clock speed is higher on the 16 core vs the 32 for applications ( or portions thereof ) that aren't multithreaded. ( Maya modeling, rigging and / or animating ) The extra cores are useful for CPU based rendering ( Arnold, Keyshot, Brazil, etc ) as they're all heavily multithreaded. So I went with a balance of the two.

Dual 1080ti's for GPU based rendering ( Blender, Octane, Redshift, etc ) and anything that can use the available CUDA cores. Refraining from 2080 cards as they are too new to be thourougly tested with the software I use.

64GB of Ram makes everthing in the Adobe CC suite giggle when launched. Upgradable to 128GB if demands ever require it.

M.2's for the performance boost vs Sata SSD.

I could have went with Xeons. . . . but why ?

Re:Throw another set of cores on the die

[Anonymous Coward]

Stop being such a pussy.

Driving at illegal speeds is great fun.

Re: Throw another set of cores on the die

[thegarbz]

Clock speed is higher on the 16 core vs the 32 for applications

That's what the core disable function is for.

Re: Throw another set of cores on the die

[Anonymous Coward]

Xeon sucks, or maybe for your use case, i9 7900X and up suck, but I wonder how much AVX-512 helps them. Seen that the i9-7940X is now available (14-core 165W TDP, vs 10 and 12 cores 140W TDP or 16 and 18 cores 165W TDP). Bit too expensive but I'd be curious about this one specifically. With power limitations, the 18-core was not any better than the 16-core. The 14-core likely can be almost as fast, or as fast, or faster than the 18-core!

32 gig unbuffered DIMMs are coming so you can upgrade to 192GB as well.

Re: Throw another set of cores on the die

[Anonymous Coward]

The 32 core may be better on the generation after that (and a drop-in CPU upgrade, except maybe for the waterblock)

I really hope 3rd generation is called...

[Anonymous Coward]

Jack.

Re:I really hope 3rd generation is called...

[Mal-2]

Did you know Jack the Ripper has been identified as... the Loch Ness Monster?

Bullshit or not?

Re: istanbul tours,istanbul excursions,daily city

[Anonymous Coward]

AMD Opteron (codename istanbul) was in 2007, let it rest in the past.

Can you imagine...

[Anonymous Coward]

A beowulf cluster of these?

Here's what I do...

[gosand]

My approach to PCs has been this:
1. Wait. Don't buy the newest tech, it's just not worth it.
2. Figure out my budget
3. Look at benchmarks - by getting older tech they are well established.
4. Get the most #3 for #2

In the last year I bought 3 'new to me' systems, one for me and two for my kids. They are used Dell Inspiron 7010s, with 8GB RAM and i5-3570 processors. They were $100 each. I picked up Nvidia GTX cards (750/460/460ti) for cheap, ~ $25 each. All said I spent less than $400 on 3 computers.

A big bump in performance for them, they play games like TF2, Fortnite, etc. and a multitude of other less-intensive games like Terraria, Mark of the Ninja, etc.
A big bump in performance for me, I run Linux and do various things like video converting, image editing, scripting, and a few games like Unreal Tournament and games from Humble Bundles.

Sure, I drooled over the new AMD offerings, but ultimately I don't need it and quite frankly can't justify the expense. This was the FIRST big-name PC I have ever owned for myself, I have always built my computers from scratch (with the exception of my first one, a 386DX-33 in 1991). But to build a new PC would take all-new EVERYTHING. (my old system was an Intel Q8400 that served me well for many years, and alas technology standards move on.) I did some video conversion timings on my old system and new, and I am seeing 6x performance increase. Could I get more out of a more expensive system? absolutely.. but it's not worth the money to me.

I guess my point is that I don't worry too much about the dizzying array of naming/numbering schemes because I know I will never ever have the fastest best system out there. Build or buy something that is good enough and carry on with your life.

ECC Memory Support Comes Free With AMD CPU's :-)

[Anonymous Coward]

Did I mention you get ECC for free with all AMD CPU's.
Fucking Intel wants you to buy Xeon at a huge price premium, or a few wimpy i3 skus, for that.
Intel has played this game on people for so long it's nuts.
People are so stupid.
You need ECC to keep from silently corrupting your shit.
It doesn't happen often, but when it does it sucks, and there's nothing you can do about it, because by the time you find it, even your backups are hosed.
And the memory only adds a few percent to the overall purchase.
Totally worth it for long term assurance.

Go Go AMD, bring down that ECC price babe.
ECC needs to be everywhere by default.
Not some relegated thing.

Remember Pentium IV ?

[DrYak]

AMD chips are also great for frying eggs.

Which is a pretty easy trick to achieve given that egg protein already start to precipitate somewhere north of 50~60C - you could achieve the same with the warm water of your faucet, try it ! Note that the egg will not have been thoroughly cooked at a high enough temperature and will not be sterilized : it might not be safe to eat due to bacterial risks.
You could do the same trick as the video with any piece of electronic more beefy that a raspberry pi

And while digging at old stuff, Intel was at the recieving end of such jokes back in the Pentium4 vs Athlon&Opteron 64 era.

(Also, since when are AMD motherboards paired with intel NICs? That seems weird to me...)