AMD Unveils Zen 2 CPU Architecture, Navi GPU Architecture and a Slew of Products (hothardware.com) 167
MojoKid writes: AMD let loose today with a number of high profile launches at the E3 2019 Expo in Los Angeles, CA. The company disclosed its full Zen 2 Ryzen 3000 series microarchitecture, which AMD claims offers an IPC uplift of 15% generation over generation, thanks to better branch prediction, higher integer throughput, and reduced effective latency to memory. Zen 2 also significantly beefs up floating point throughput with double the FP performance of the previous generation. AMD also announced a 16-core/32-thread variant, dubbed Ryzen 3950X, that drops at $750 -- a full $950 cheaper than a similar spec 16-core Intel Core i9-9960X. On the graphics side, AMD's RDNA architecture in Navi will power the company's new Radeon RX 5700 series, which is said to offer competitive performance to NVIDIA's GeForce RTX 2070 and 2060 series. The Navi-based GPU at the heart of the upcoming Radeon RX 5700 series is manufactured on TSMC's 7nm process node and features GDDR6 memory, along with PCI Express 4.0 interface support. Versus AMD's previous generation GCN (Graphics Core Next) architecture, RDNA delivers more than 50% better performance-per-watt and 25% better overall performance. Greater than 50% of that improvement comes from architecture optimizations according to AMD; the GPU also gets a boost from its 7nm process and frequency gains. Radeon RX 5700 and 5700 XT cards will be available in market on July 7th, along with AMD Ryzen 3000 chips, but pricing hasn't been established yet for the Radeon GPUs.
Rocking my first AMD card since the 1650x (Score:4, Interesting)
AMD seems to have gotten their act together with drivers, at least as much as nVidia. If they can get power consumption under control they might finally get back some market share.
Re: (Score:2)
The RX580 was pretty competitive on power consumption, compared to the equivalent Nvidia cards. The current desktop Ryzen and Threadripper CPUs are very competitive too, although less so for laptops.
I haven't tested it with a Wattman or anything (Score:2)
Am I totally off base here? Happy to be proven wrong
Re: (Score:2)
I had a quick look at some reviews and they seem pretty similar. I have a 480 which is the same architecture but less power efficient and wouldn't say it's particularly bad either.
Re: (Score:2)
"I'm also surprised how well older games run on it."
i don't understand, is there a problem with old games on modern gpu's?
i would assume they only run better as the tech in the gpu advances.
Re:Rocking my first AMD card since the 1650x (Score:4, Interesting)
Some old games rely on depreciated APIs or functions. The newer hardware may have to emulate them in software.
A few old games run like utter ass on my 1080Ti.
I remember going from my 1650x (Score:2)
These days image quality is about the same though, so it's mostly price/performance.
Re: (Score:3)
That's what 7nm Navi does: gets power consumption under control. Well, and more, but that's the big one for me.
Re: (Score:2)
Re: (Score:2)
Used cards aren't that much lower on sale than new ones nowadays. It was expected that we'd have a firesale after the recent crypto crash, but that never happened. Prices stayed at about -20% to -30% compared to new cards. Definitely not as low as 100USD for RX580. That's a steal.
Re: (Score:2)
First 580 that pops up on ebay is $77 (used). You could have known this for yourself instead of just spouting.
Re: (Score:2)
They are now because both nvidia and AMD pushed another gen through so prices dropped. Those are old gen cards that are going for 130-ish new.
Re: (Score:2)
So, still spouting.
Re: (Score:2)
I'm sorry, I assumed intellect sufficient to understand that time is linear on your part. I stand corrected.
Re: How did you pay 100 USD for it?!?! (Score:2)
Re: How did you pay 100 USD for it?!?! (Score:2)
Re: How did you pay 100 USD for it?!?! (Score:2)
AMD killing it lately (Score:5, Insightful)
Remember when AMD was stuck on Bulldozer and related cores? Intel does, because they basically didn't advance for that entire period.
Now AMD is back and killing it and Intel is scrambling again. Competition is heating up and I got to buy a 4-core 8-thread chip that draws peanuts at idle and 65w all-out for $150, AND it includes graphics equivalent to the low end discreet market.
I missed this rapid progress these last 5 years. Fuck you Intel for riding "good enough" forever.
Re:AMD killing it lately (Score:4, Informative)
Re: (Score:2)
Agreed. Turning AMD around was a bloody miracle! The fact that they are competitive from both a finance AND performance POV with ThreadRipper is almost mind blowing. My next rig is going to be a ThreadRipper one.
Kudos to Lisa for doing the next to impossible. Almost everyone thought AMD was a goner when they sold their FABS but that thankfully has worked out for them.
Intel has (historically) almost always been faster but you sure pay through the nose for it! Glad to see AMD giving customers what they want.
Re: (Score:2)
when Intel isn't "cheating" on benchmarks by detecting they are running on non-Intel chips and defaulting to slower code!
Or rather, running micro-specific code when an Intel proc is detected.
It's a subtle, but important difference.
It's definitely a questionable practice, but I think it does the argument no good by framing it disingenuously.
Re: (Score:2, Informative)
Not "rather". What he says is exactly what happened [agner.org].
If the CPU is not from Intel then, in most cases, it will run the slowest possible version of the code, even if the CPU is fully compatible with a better version.
Re: (Score:2)
Re: (Score:3, Funny)
Re: (Score:2)
Did you just assume Lisa's... nevermind.
intel hooked the mac pro so they can make bank on (Score:2)
intel hooked the mac pro so they can make bank with the mark up on that.
Re: (Score:2)
Re: (Score:2)
intel hooked the mac pro so they can make bank with the mark up on that.
Apple has what percentage of the market? How many of those machines will they be selling? Intel will make jack with the number of sales involved with that. There, FTFY.
Re: (Score:2)
intel hooked the mac pro so they can make bank with the mark up on that.
At $6,000 for the low end model, how many do you think Apple is going to sell? The Mac Pro (even before the 2013 model disaster) has always been a niche product. Even more so now that they have given the finger to the pro-sumer market.
Re: (Score:3)
It is awfully familiar...
20 years ago Intel was banking on Itanium and Netburst heavily. Both were highly ill advised moves. AMD came out with a sleek backwards compatible 64-bit architecture with a mainstream NUMA architecture for servers. AMD *should* have been on top of the world, and only curtailed in the market by unfair Intel practices and generally better marketing than AMD could field as well.
Then AMD went Bulldozer and Intel went Nehalem and fortunes reversed dramatically, as AMD made the dumb m
Re: (Score:3)
https://www.newegg.com/core-i7... [newegg.com]
Also 65w, but with the MUCH slower 630 graphics?
$323.59?
https://www.newegg.com/intel-c... [newegg.com]
The 35w part, but the clock is reduced to 2.9ghz with turbo 3.6ghz. I assume the graphics clock is reduced as well.
$349.96
https://www.newegg.com/amd-ryz... [newegg.com]
Ryzen 2400g
$139.99
Re: (Score:2)
Re: (Score:2)
If only Intel would fix their current chips for the threats of yesterday.
Re: (Score:2)
They took a silly road. Sacrificing performance for perceived security that is actually completely irrelevant outside a very narrow use case.
Re: (Score:2)
Wonder if (Score:2)
Re: (Score:2)
Possibly. I think I read somewhere that Ryzen supports ECC RAM, but it is not part of the official validation testsuite. Though if Apple wanted a workstation class CPU with official ECC support I'm sure AMD could produce something real quick.
But I'm guessing Apple is happy to be seen as "XEON Premium" as long as the performance is OK. They might also have a (possibly unwritten) exclusivity agreement with Intel, for the time being.
Spectre fixes? (Score:2)
This is neat and all but did they fix the issues related to speculative execution, specifically the Spectre class of vulnerabilities? Intel seems to have put off fixing anything in their microarchitecture for a full decade and I was hoping AMD would actually respond faster.
Re:Spectre fixes? (Score:2)
on the second line of OP, I read : "...uplift of 15% ... thanks to better branch prediction..."
If this is not speculative execution...
Re:Spectre fixes? (Score:4, Informative)
It's not the speculation which is the problem, it's what you're allowed to do with it. Intel's problem was that they executed code speculatively before even checking if you were allowed to perform the action, and then tried to win the race to remove the result ex post facto before anyone could read it. IOW, they let you read and copy all the secrets and then tried to shoot you and destroy the copies before you could get away or hand over the information to someone else.
AMD CPUs first check if the operation is permissible before it executes.
Re: (Score:2)
It's more like mission impossible bad as in the cost of all the counter measures to this are utterly ridiculous, I mean do you want to inoculate your janitorial staff against sedatives??!
Then why is AMD basically immune to this kind of attack?
Your rebuttal makes no sense, on its face. You describe something that both CPU's should be vulnerable to, but for some reason only one is. I'm not saying your post parent is correct (I honestly have no clue), I'm just saying their version of what is going on is actually consistent with the reality that Intel is vulnerable to these attacks, while AMD is not.
That makes me think it's a lot more likely they are correct than you are.
Re: (Score:2)
Why? What secrets are you hiding from the NSA that makes you worried about something like Spectre? And for which computer? Hopefully not the one you stupidly connected to the internet to make this comment.
Why don't you take your security seriously? Are you in management chasing buzzwords? "Ooooh Spectre, sounds scary, must be bad"
Please provide a bit of thought into your risk. In the meantime I'm hoping AMD will start shipping faster CPUs that are vulnerable to Meltdown, because I much like everyone else wh
Re: (Score:2)
You might be fine with someone else having control over your computer, but then you'd be an idiot. Are you an idiot?
First you need to prove to me that Spectre / Meltdown allows arbitrary control to be taken over a PC. Based on all evidence and all published documents it's the reverse where a significant degree of control must be had already be given up to a 3rd party.
I don't know about you, but my home security involves a locked door, not relying on sitting next to my valuables with a gun to kill someone already in my house.
Re: (Score:2)
Intel shilling? I suggest you read my posts to see how stupid your comment actually is. I am a 100% AMD fan. They are the absolute kings of performance per dollar and have been for the past 2 years and I wouldn't recommend anything else.
I just wish that AMD and Intel were equally susceptible to Meltdown and Spectre, as the result would be I'd own an even faster AMD machine than I do now.
Re: (Score:3)
More multi-threaded applications (Score:2)
Already have a Radeon 5770... (Score:2)
...so I'm good, right? I mean, it's 70 better than a 5700, right?
Re: (Score:2)
It should be as powerful as ~2.787 times the nVidia RTX 2070, so you should be okay for a while.
Re: (Score:2)
still steps behind intel and nvidia. but keep at it maybe intel will through yah a bone one day
Not really at all behind Intel. In fact, they're pushing faster and farther than intel currently.
Re: (Score:2)
If by behind Intel you mean way way ahead of Intel, then yes, definitely. AMD is winning on absolute single threaded performance, multithreaded performance, cost and power efficiency, at least based on what we know so far (third party benchmarks will be more definitive), plus they will be first to market with PCIe 4 support (which Intel hasn't even released a roadmap for yet). When 3rd gen Ryzen launches in less than a month, Intel will be so comically uncompetitive in CPUs on every front, you'll wonder how
Re: (Score:2)
Navi never was aimed at 2080+, however don't be surprised if it does end up giving those NVidia flagships a serious scare.
Navi is just going to kill everything below 2080. I bet Navi raytraces better than RTX 2070.
Re: yawn (Score:2)
I bet Navi raytraces better than RTX 2070.
Now that's fucking retarded.
Re: (Score:2)
Spoken as a random idiot with zero understanding of the technology. Watch this [youtube.com] and understand that Nvidia's RTX scam is doomed.
Re: (Score:2)
Though that isn't to say that what CryTek pulled off isn't impressive. [techspot.com]
Re: (Score:2)
Competing with the halo parts is still important. It helps get sales lower down. Just like the corvette on the lot gets people to buy a minivan, the 2080ti gets people to buy 2070s. AMD needs to release a 5700XTX, not because they will move many units but because it will help sell 5700XTs.
Re: (Score:2)
The Radeon VII already meets or beats NVidia's halo parts, except the ultra high end costing nearly twice as much. Obviously, it is only a matter of when AMD gets around to it, to kick the 2080 ti as well. See, Nvidia inexplicably took a pass on 7nm and now they will pay for it.
Re: (Score:2)
Maybe with their smaller process they can finally match their competitors with 50% larger processes? That'd be cool.
Still good bargain parts though, I won't lie.
Re: (Score:2)
The problem is that it looks like the Navi GPUs still use 2080+ levels of energy to compete with 2070, which isn't the best news for power bills, noise, and related issues.
In fact, if Navi did compete with 2070 with comparable power envelope, you would bet they'd have a competitor to the 2080s...
Also, this ignores the allegedly impending 20 'super' coming out from nvidia.
While on the CPU AMD is doing great performance and price performance and performance per what, on the GPU side they are still relegated t
Re: (Score:2)
it looks like the Navi GPUs still use 2080+ levels of energy to compete with 2070
You haven't been paying attention. Read the specs closer.
Re: (Score:2)
AMD is winning on absolute single threaded performance
Wait, what?
I missed that. Do the new procs kick Intel ass? Because as of the previous generation, at least, the i9-9900K still eats AMDs lunch for N cores, where N I'm not trolling, I'm really just asking if they've finally turned the tide.
Re: (Score:3)
* Where N is less than or equal to the amount of cores in the Intel proc.
Re: yawn (Score:2)
Re: (Score:2)
and then meltdown etc. rendered all of that investment moot
Meltdown has done nothing of the sought. There's no need to gimp the performance of a processor because you're worried about being struck by a meteor.
Re: (Score:3)
That is pretty much all Intel was winning at and it was a narrow lead, certainly not one within any reasonable metric of expense. The difference was so marginal that slight tuning variation or variation of cpu off the line could eat it, only at the top end and dramatically higher priced. It was definitely eaten if you patched.
If the specs on the new architecture are even half true then yes, Intel is officially beaten across the board except in some narrow laptop edge cases with onboard graphics.
Re: (Score:2)
That is pretty much all Intel was winning at and it was a narrow lead
If by narrow, you mean around 30%, then sure.
certainly not one within any reasonable metric of expense
I'm having difficulty parsing that. Are you saying not by any metric that matters? If so, you're talking out your ass.
The difference was so marginal that slight tuning variation or variation of cpu off the line could eat it
No. Sorry. You're speaking from the armchair. I do large scale datacenter deployments, and there's a reason we're *still* doing Intel. Per-core performance is important for our workloads. Hell, it is for most peoples' workloads.
It was definitely eaten if you patched.
Nope. With older architectures- absolutely. But that no longer applies.
If the specs on the new architecture are even half true then yes, Intel is officially beaten across the board except in some narrow laptop edge cases with onboard graphics.
I'm excited to see if this pans out. We'd like to
Re: (Score:2)
"If by narrow, you mean around 30%, then sure."
It seems like you've been reading reviews sponsored by Intel. Try 10-20% prepatch in the best case.
"I'm having difficulty parsing that. Are you saying not by any metric that matters? "
No, reading doesn't seem to be your strong suit. I said expense not "matters." An intel cpu with a margin single thread performance and otherwise inferior performance would cost you dramatically more, even double.
"Nope. With older architectures- absolutely. But that no longer appl
Re: (Score:2)
It seems like you've been reading reviews sponsored by Intel. Try 10-20% prepatch in the best case.
That's absolute nonsense. 10-20% in the best case for *AMD*.
No, reading doesn't seem to be your strong suit. I said expense not "matters."
LOL.
You said: "That is pretty much all Intel was winning at and it was a narrow lead, certainly not one within any reasonable metric of expense."
I was trying to give you the benefit of the doubt by parsing it into something that made sense, since that sentence right there- does not.
What the fuck is a "reasonable metric of expense" for a lead in performance? One potato per onion, if you ask me.
My reading isn't the issue here. I suspect English i
Re:yawn (Score:4, Interesting)
"I'm excited to see if this pans out. We'd like to leverage the obscene core count/price ratio offered by AMD. I just can't in good conscience drop our single threaded performance when we have enough task loads dependent on per-thread execution speed."
Here at least we can come together. You can dispute AMD EPYC being the top dog but we all want out from under Intel. They haven't exactly been a benevolent steward of the CPU market the last decade. Opteron sucks and a lot of people made the mistake of buying into the too good to be true cheap price and are afraid of being burned again. EPYC is a whole different beast and the next gen will settle the case. The DC actually moves slow though so they'll need to beat Intel on more than a couple releases if they want to win over the DC.
Re: (Score:2)
You can dispute AMD EPYC being the top dog but we all want out from under Intel.
Dispute? There's nothing to dispute. Xeons are the top dog. That's just a fact. Server churn in our datacenters is about yearly, and yet AMDs still haven't gained an inch of ground.
The only area where EPYCs matter in the DC, is pure multithreaded performance- which is phenomenal. But not enough.
The world's datacenter workload is not embarrassingly parallel. It just isn't. The only people who think it is, are people who don't fucking do this work.
Re: (Score:2)
AMD is winning on absolute single threaded performance
Wait, what? I missed that. Do the new procs kick Intel ass?
Yes. The ~10% increase in clock speeds with 3rd gen Ryzen, plus AMD's claims of 15% better IPC compared to 2nd gen Ryzen is enough to comfortably put them in the lead.
Now, we don't have third party benchmarks yet, but the the single threaded performance gap was only in the 10-15% range (though with fairly large swings in both directions, depending on the benchmark) for 9900k vs 2700x. So even in the worst case scenario where that IPC improvement is close to zero (and we have no reason to expect this), they
Re: (Score:2)
So, beyond that quibble- another 10% clock on the 2700X (probably it's single largest shortcoming against the Intel) and a 15% improvement in IPC should definitely at the least put the new AMD 8/16 on equal footing with the 9900K.
And of course, I'm definitely not trying to start an argument on performance per dollar, where there's no question who holds the crown.
Re: (Score:2)
AMD is winning on absolute single threaded performance
Nope. Ryzen 2700 only beats out my i7-3770K by 1% in single core [userbenchmark.com] performance. It can only beat a comparable i5-8600 (not even i7) in multicore [userbenchmark.com] performance.
I'm looking forward to seeing what the Zen2 Ryzens can do, but it will probably be months after their general market release before we'll get a clear picture.
Re: (Score:3)
Re: (Score:2)
AMD is winning on absolute single threaded performance
Nope. Ryzen 2700 only beats out my i7-3770K by 1% in single core [userbenchmark.com] performance.
Based on AMD's numbers, the 3800x will beat the 2700x by ~25% on single threaded performance. We'll have third party benchmarks to confirm when it launches July 7th.
Re: (Score:2)
definitely. AMD is winning on absolute single threaded performance, multithreaded performance
Let me bring you back to reality. In the past 3 processor generations AMD has not once outperformed Intel on single threaded performance. AMD has not once outperformed Intel on multithreaded performance.
They are however damn good in the cost and power categories for relative performance.
PCIe 4.0 is completely uninteresting and ignores something that AMD IS good at: They provide far more PCIe interfaces to their CPUs than comparable intels and for threadripper provide free CPU RAID support (something that ne
Re: (Score:2)
definitely. AMD is winning on absolute single threaded performance, multithreaded performance
In the past 3 processor generations AMD has not once outperformed Intel on single threaded performance.
True only up to 2nd gen Ryzen. 3rd gen Ryzen is beating Intel here, at least based on what we've seen so far. Third party benchmarks are coming in under a month, when the CPUs launch.
AMD has not once outperformed Intel on multithreaded performance.
Hilariously untrue. Have you forgotten how long Intel was pushing quad cores (and later, 6 cores) as their top consumer level CPU vs 8 core Ryzens? Intel was getting absolutely destroyed on multithreaded performance until they launched the 9900k late last year.
Also, if you want to talk about "ever", let's talk about the Athlon
Re: (Score:2)
If Intel comes out with a 7 GHz boutique chip with carbon nanotube cooling system priced at $1 million each, in your opinion would Intel's crown size increase by another 50%?
Because I'm over here in the podunk world of only comparing like with like, while completely ignoring the Spendo Silverback Supercustom SKUs.
Re: (Score:2)
Re: (Score:2)
Except that with Ice Lake, Intel is catching up on graphics while increasing their lead on computing. The half hearted APU update we're getting for 2019 is a yawn. Perhaps we will see a 7nm APU from AMD in 2020 that will finally get them into the laptop game.
But AMD is ruling the desktop right now. Intel chose to put its first mainstream 10nm effort into laptops to solidify its position there rather than immediately releasing an answer to the Ryzen 3000 series. They won't be able to coast on momentum there
Re: (Score:2)
Indeed since AMD offerings are outperforming them there as well.
Re: (Score:2)
Looks to me like matched or beat Intel in all but a couple of boutique market sectors involving ridiculously cherry picked parts, stupidly overpowered to the point of needing industrial cooling. And completely clobbered Intel in value, there is not a shred of doubt about that.
Posted from my el classico 1700 :)
Re: (Score:2)
AMD is selling a 12 core for a fraction of the price Intel is. They are winning in just about every metric right now. The Intel chips that are faster are way more expensive, which means they aren't competing. And the ones that are competing on price are slower and less feature packed. The streaming demo was very impressive. Any midrange and up cpu today is plenty good for gaming but the 3900X being able to game at 1440p and simultaneously run OBS in quality mode with 98% frames streamed.....while a 9900k ma
Re:yawn (Score:4, Insightful)
When you consider that Intel gets much of its performance advantage from an insecure architecture and asks much higher prices, not really. But what can you expect from an AC shill.
Re: (Score:2)
When you consider that Intel gets much of its performance advantage from an insecure architecture and asks much higher prices, not really. But what can you expect from an AC shill.
Maybe some advice to AMD. Copy Intel. No really, do it. Spectre and Meltdown don't impact me. Give me the performance!
Re: (Score:2)
Re: (Score:2)
Maybe because Apple and I-stuff are well on their way to irrelevance.
Re:Spectres fixes? (Score:5, Interesting)
They were much less susceptible than Intel already before, i.e. attack execution is much, much harder. They have some fixes in there now (before was just better architecture). Whether the result is a "complete" fix, i.e. the attack is either not doable anymore or so much effort that it does not make sense to use it, or whether they just made it even harder is unclear. AFAIK, most of the Spectre-class attacks were only ever demonstrated on Intel and some may have been practically infeasible on AMD even before.
As AMD has now a superior DC offering, I expect people will try to overcome the higher hurdles to see whether the attacks work or not. But this may take a while. These attacks are all hard pretty to do, even the easier ones on Intel.
Re: (Score:2)
There are a handful of the 11 plus variants (all affect Intel) there are about 4 that affect AMD. But side-channel attacks aren't going away, now that there is a exploit threat model the exploits will likely keep coming as researchers continue to probe.. But as you say, if it follows precedent the AMD chips will be far less vulnerable than the Intel ones because AMD tried to design secure, where Intel cut corners to gain a little extra speed.
Re: (Score:2)
Re: (Score:2)
The 3000 for laptops are actually 2000.
The laptop (and desktop APUs) are a generation behind.
Re: (Score:2)
Did they fix the Spectre vulnerabilities in silicon now?
Who cares?
No really who cares? Show me the person so poor at risk assessment that they think that Spectre is some great vulnerability and yet will still dare to risk connecting their PC to the internet.
Re: (Score:2)
Why this fixation on a fix in silicon when you can fix many of them in software with minimal performance degregation (on AMD at least)? It's not that fixing it in silicon will magically increase the performance either. If you are really worried just download the latest checker script from https://github.com/speed47/spe... [github.com] , running it on even a first gen Ryzen tells that of the 12 different vulnerabilities 9 of them are "(your CPU vendor reported your CPU model as not vulnerable)" and the remaining 3 are fi
Re: (Score:2)
They weren't vulnerable to most of them in the first place.
Re: (Score:2)
"they" in context is AMD not Intel, Ryzen 3000-series are AMD chips. "They" would not be shouting from the rooftops at fixing Spectre vulnerabilities in silicon because "they" weren't vulnerable to most of them in the first place. Intel is heavily impacted across the board.
Re: (Score:2)
Anybody else think that's a sweetheart deal for a 16-core 32-thread consumer level CPU?
I do want. I do. Stuck with intel's 8T since forever now :(. I've got way too much workload that could benefit from a mess of threads.
Oh yeah.
Re: (Score:2)
AMD put out their mainstream GPU first, they will fill in the budget parts pretty soon. 1660 is toast.
Re: (Score:2)
1660 can do ray tracing too, just like 10xx parts can. It just doesn't have the denoiser parts.
And we have no idea what AMD's implementation will actually be. Let's wait for the actual part to be benchmarked before we start making conclusions.
P.S. Also why the fuck does AMD still use blower designs on their GPU coolers? That cripples their thermals and as a result, performance and longevity.
Re: (Score:2)
we have no idea what AMD's implementation will actually be
You can bet it's in an advanced state and will be released only when it's actually usable, unlike Nvidia's lame ass puddle tracing RTX decelerator. Nobody needs to drink Nvidia's RTX koolaid, see here. [youtube.com]
Yah, Navi is a dagger aimed straight at the heart of 1660. And 2080 might well go down as collateral damage.
Re: (Score:2)
Try using a non-blower card in a mini PC case such as the A4-SFX. You will toast your GPU in a matter of minutes/hours. With a blower design, the heat exits at the back.
Re: (Score:2)
There are people who have done just that on youtube who do this sort of testing professionally. Outcome is that it's either similar or better outcome than that of blower design.
And when you're selling parts that are going to go for over 200USD, it's not unreasonable to assume that case will have a single exhaust fan.