Question AMD preps Ryzen 9 3900XT, Ryzen 7 3800XT, Ryzen 5 3600XT for June 16 ....TweakTown

[JPB]

AMD rumored to refresh Matisse CPUs with Ryzen 7 3850X and 3750X

A big juicy rumor has just hit the internet, with it teasing that AMD would be refreshing its current-gen Ryzen 3000 series CPUs with two new entries.



These two new processors are the Ryzen 7 3850X and the Ryzen 7 3750X, which would succeed the Ryzen 7 3800X and Ryzen 7 3700X processors that are currently on the market. The tweet comes from 'HXL' who published the SKUs of the two new processors and made sure to tell the world they are "Not APU".

The tweet itself was a reply his own tweet of a 'Big day coming...' and that the release would be on 6/16 which after that AMD would launch on 7/7. An interesting timeline of release for the Matisse refresh, as 7/7 was the date that the Zen 2-based Ryzen 3000 series launched in 2019.

If you didn't believe just that single tweet, then how about GIGABYTE kinda confirming the Matisse Refresh in their B550 presentation slide, as noted by VideoCardz. The GIGABYTE motherboards roadmap for support shows "Matisse & Refresh",

"Vermeer", and "Renoir" -- all of which aren't released yet.

AMD seems to be responding to the just-launched Intel Core i9-10900K and Core i7-10700K processors, as these new Ryzen 7 3850X and Ryzen 7 3750X processors would directly compete against Intel's new 10th Gen CPU offerings.

AMD preps Ryzen 9 3900XT, Ryzen 7 3800XT, Ryzen 5 3600XT for June 16

We first heard wind of AMD's upcoming Matisse Refresh in some leaked news barely 24 hours ago, but now we have some even juicier stuff to talk about: the Ryzen 9 3900XT, Ryzen 7 3800XT, Ryzen 5 3600XT.

That's right, according to our friends at Wccftech, AMD is about to launch the Matisse Refresh with the Ryzen 9 3900XT, Ryzen 7 3800XT, Ryzen 5 3600XT processors. The new Ryzen 3000 series CPUs would adopt the "XT" branding, something that Radeon fans will remember and love to see back in the ring.

The new rumor has AMD replacing the Ryzen 9 3900X, Ryzen 7 3800X and Ryzen 5 3600X with the new Ryzen 9 3900XT, Ryzen 7 3800XT, Ryzen 5 3600XT processors -- but increasing the CPU clocks, having enhanced overclocking support, and dropping the price on the current-gen CPUs that they replace.

Ryzen 9 3900XT - 12C/24T + faster clocks + enhanced OC support
Ryzen 7 3800XT - 8C/16T + faster clocks + enhanced OC support
Ryzen 5 3600XT - 4C/8T + faster clocks + enhanced OC support

These new CPUs would compete directly against Intel's latest Comet Lake processors, in the just-launched Core i9-10900K, Core i7-10700K, and Core i5-10600K.

 

[piokos]

Unlikely. Intel isn't using Tiger Lake in enough product segments for it to matter that much.

Intel is going to use Tiger Lake in their mobile segment, which is larger than AMD as a whole. So I'm not sure where you're going with this "not enough to matter".

 

[Hitman928]

That's gross margin. Possible, given 46% total gross margin. Can you link the source?

AMD's profit margin was 9%.
And it was a very good quarter - they actually expect the rest of the year to be worse.

By comparison, Intel's profit margin for PC CPUs is 35% and for data center chips it's around 40%. So in their case Xeons are more profitable, but again: their reference pricing is way above AMD's.

Yes gross margin, but you said AMD was selling them near cost or even lower. Gross margin above 50% says they aren't selling anywhere near cost.

I'd have to go through all of the last few earnings calls to find the exact quote which I don't have time to do right now. A quick google gives me lots of supporting quotes though.

From AMD's 2019 full year comments

Gross margin was 43 percent compared to 38 percent and non-GAAP gross margin was 43 percent compared to 39 percent in the prior year. Gross margin expansion was primarily driven by Ryzen and EPYC products.

Press Releases

Browse AMD’s company-wide and financial press releases.

ir.amd.com

From AMD's 1Q20 comments

We executed well in the first quarter, navigating the challenging environment to deliver 40 percent year-over-year revenue growth and significant gross margin expansion driven by our Ryzen and EPYC processors

Press Releases

Browse AMD’s company-wide and financial press releases.

ir.amd.com

Anandtech comments on AMD earnings back in 3Q18

This was offset though by increase server sales with EPYC, and the additional EPYC sales also helped with the margins, thanks to enterprise offering much better returns than consumer plays.

AMD Announces Q3 FY 2018 Earnings: Improving Margins, But Dogged By Crypto Decline

www.anandtech.com

AMD is also projecting flat (actually 2pp higher) gross margin for the year despite ramp up of semi-custom revenue from consoles which is notoriously low margin. So they'll have to continue to increase margin in the rest of the business (see increased Epyc sales) to do that.

 

[Mopetar]

Even if we see some refreshed Zen2 parts AMD won’t delay desktop Zen3. Eventually Intel will get out of the rut they’ve been in and will be more of a serious contender. AMD has nothing to gain from sitting on their own hands.

If they have supply shortages they can always release the high end of the stack first. Halo products always generate the most buzz and hype even if they don’t sell anything like the mainstream parts do. They’ll also get some people to wait to buy AMD in the future if it’s only a few more months.

 

[Kuiva maa]

@Makaveli

Unless they're deliberately going to delay Zen3. Right now its release is probably in October (which is kind of late but oh well).

Zen 1 came out in March 2017. Zen+ which was just a refresh in April 2018 and Zen 2 in July 2019. AMD is on a 13-15 month cadence between generational releases so an October 2020 release for Zen 3 would actually be right on schedule (15 months). I expect it to come later due to covid actually and this refresh could act as a bridge.

 

[moinmoin]

Besides all of the above, they have also stated publicly multiple times that zen 3 will come "late this year".

"later this year" actually. "late this year" was specifically for Milan.

3900x was $499, 3950x was $749. 3950x carried that premium because it was the first time a 16c had happened on desktop. Now they may not lower it, but if they are going to leave open the prospect of a higher core count on a desktop socket . . .

The premium will be reduced once there's something above it so no longer deserves the premium, spacing the products evenly in the available price range. E.g. Threadripper 3 core by core actually got more expensive compared to previous Threadripper gens so you are being quite optimistic expecting lower prices with the new gen expected to introduce significant performance improvements.

 

[jpiniero]

There probally is some amount of trying to coax the hardcore early adopters to buy a more expensive model than they would have if there were cheaper Vermeer models available initially.

 

[Ajay]

Uh

3900x was $499, 3950x was $749. 3950x carried that premium because it was the first time a 16c had happened on desktop. Now they may not lower it, but if they are going to leave open the prospect of a higher core count on a desktop socket . . .

Well, I'd be surprised if we get more than 16 cores on the AM4 socket just because of the routing problems with Zen2; never mind Zen3. Also, since there is not going to be any significant shrink on 7N (whichever one), there just isn't enough physical space (more IPC == more area). I expect AM5 to be a larger socket (among other things); with the shrink from 5N, maybe we'll see more cores on the 'desktop' platform.

 

[AnandThenMan]

Did you know your performance changes depending on the profit margin the CPU maker has at the time? True story.

 

[kawi6rr]

This would be a great upgrade from my 2600x, crossing fingers.

 

[piokos]

Yes gross margin, but you said AMD was selling them near cost or even lower. Gross margin above 50% says they aren't selling anywhere near cost.

Gross margin includes only direct manufacturing costs. In case of AMD that's pretty much what they pay TSMC.

I said "break-even" which means with no profit. You're looking at the wrong margin.

I'd have to go through all of the last few earnings calls to find the exact quote which I don't have time to do right now. A quick google gives me lots of supporting quotes though.

You have time to write posts with this kind of claims, but you don't have time to support them with a source.
No source => it didn't happen.

From AMD's 2019 full year comments

Press Releases

Browse AMD’s company-wide and financial press releases.

ir.amd.com

From AMD's 1Q20 comments

Press Releases

Browse AMD’s company-wide and financial press releases.

ir.amd.com

Anandtech comments on AMD earnings back in 3Q18

AMD Announces Q3 FY 2018 Earnings: Improving Margins, But Dogged By Crypto Decline

www.anandtech.com

I know how where to find financial statements. You don't have to post links.

Is there anything here that says or at least suggests gross margin on EPYC above 50%? Because that's what you said.

AMD is also projecting flat (actually 2pp higher) gross margin for the year despite ramp up of semi-custom revenue from consoles which is notoriously low margin. So they'll have to continue to increase margin in the rest of the business (see increased Epyc sales) to do that.

Or console stuff isn't low-margin anymore. Or something else.

 

[Hitman928]

Gross margin includes only direct manufacturing costs. In case of AMD that's pretty much what they pay TSMC.

I said "break-even" which means with no profit. You're looking at the wrong margin.

Fair enough. I think most people would consider break even to be in terms of production costs, but I get that this wasn't what you meant. With that said, it's still a very bold claim and you'd need to break down R&D across both Ryzen, Threadripper, and Epyc because they all use the same chiplet design. You also need to account for ongoing costs for LTS of products that are still a part of costs for AMD as well as the number of legacy products which are low margin that AMD is still selling. You also have to break out debt interest and reduction calculations as well. You are the one claiming they are selling Epyc at break-even when no one else seems to hold that opinion, so if you can back that up with sources or calculations based upon their revenue, costs, and commitments, I'd love to see it.

You have time to write posts with this kind of claims, but you don't have time to support them with a source.
No source => it didn't happen.

It took me about 5 minutes to write that last post and less for this. It could potentially take me an hour to find the quote trying to go through multiple earning call transcripts.

I know how where to find financial statements. You don't have to post links.

Is there anything here that says or at least suggests gross margin on EPYC above 50%? Because that's what you said.

Simple math. If gross margin was at 40% and is driven to mid 40% thanks to Ryzen and Epyc when Epyc is less than 10% of revenue, the math tells you that Epyc has to be selling for above 50% gross margin. Unless you think AMD is lying to investors about this.

Or console stuff isn't low-margin anymore. Or something else.

Console stuff is still low margin. AMD addressed this as well. When I have more time I can pick out the actual quotes for you. If you think it's something else, then please enlighten us as to what could be driving it.

 

[darkswordsman17]

Gross margin includes only direct manufacturing costs. In case of AMD that's pretty much what they pay TSMC.

I said "break-even" which means with no profit. You're looking at the wrong margin.

You have time to write posts with this kind of claims, but you don't have time to support them with a source.
No source => it didn't happen.

I know how where to find financial statements. You don't have to post links.

Is there anything here that says or at least suggests gross margin on EPYC above 50%? Because that's what you said.

Or console stuff isn't low-margin anymore. Or something else.

Well this makes sense.

"No source so why did you bother."
*Looks at sources.*
"I didn't need you to post sources."

OEMs would be fine with selling Matisse Refresh for gaming desktops.



I imagine the 3600X/3800X/3900X will be discontinued when the XT parts launch, and the 3950X when Vermeer launches.

Yeah. I'd guess OEM Zen 3 won't happen til next year anyway. It even kinda works out well, as AMD can show off OEM Zen 3 products at CES, have them launch early spring, and then if there's a Zen 3+ those could be ready for fall.

Why not just adjust price? They roll these out at the same price as the X variants had, while they rework the price structure. It lets them keep their premium status, while also dropping prices.

I still have a personal hunch that this "refresh" isn't what we'd normally consider a refresh (like Zen+ was), but is more like them just having binned a decent amount of chips that could clock higher, and so they can roll them out as fresh SKUs to tide people over til Zen 3, and to keep stealing thunder from Intel.

 

[jpiniero]

Why not just adjust price? They roll these out at the same price as the X variants had, while they rework the price structure. It lets them keep their premium status, while also dropping prices.

Too many products, especially depending on when Renoir Desktop is also available.

 

[GodisanAtheist]

AMD just clearing out excess Zen2 inventory prior to the launch of Zen 3.

As has been mentioned in this thread, would not be surprised if Zen 3 initially comes to us in Epyc and Threadripper form, then works its way down into mainstream desktop markets in early 2021.

 

[Makaveli]

AMD just clearing out excess Zen2 inventory prior to the launch of Zen 3.

As has been mentioned in this thread, would not be surprised if Zen 3 initially comes to us in Epyc and Threadripper form, then works its way down into mainstream desktop markets in early 2021.

Agreed we will see Zen 3 in Epyc and TR first. The consumer version we may still see before the end of the year but it will be the 12/16 core parts only.

 

[DrMrLordX]

OEMs would be fine with selling Matisse Refresh for gaming desktops.

I guess? AMD is facing an uphill battle with OEMs, and Zen3 is likely a more-compelling product. OEMs have been stuck selling the same Intel core to customers for years. Do you really imagine they're eagre for AMD to treat them the same way?

I imagine the 3600X/3800X/3900X will be discontinued when the XT parts launch, and the 3950X when Vermeer launches.

My point was that if they raise prices, their opportunity to snap up marketshare will be wasted.

Intel is going to use Tiger Lake in their mobile segment, which is larger than AMD as a whole. So I'm not sure where you're going with this "not enough to matter".

1). We still don't know total volume. Tiger Lake is sharing that space with 14nm parts.
2). Intel has many other market segments where Tiger Lake will not be sold, which was my point. A good deal of their intended volume will still be 14nm.

The premium will be reduced once there's something above it

That was why I mentioned the possibility of lowering the price point for the 16c part (leaving open the possibility of a higher-core-count part). If AMD is saving that for Zen4, then the 16c part should keep to the $749 MSRP.

 

[DisEnchantment]

If Zen3 desktop is indeed coming in 2021, AMD could have escaped being pestered about BIOS compatibility with existing boards 🤭

 

[Makaveli]

If Zen3 desktop is indeed coming in 2021, AMD could have escaped being pestered about BIOS compatibility with existing boards 🤭

nah they would still bitch about it.

lol people are people.

 

[StinkyPinky]

It's possible they are just getting really good yields and want to bump up the mhz slightly to offset Intel. Interesting times. Thank god for Ryzen otherwise we'd probably (at best) have a 6 core 10700K as the flagship desktop consumer cpu at $100 more. We all know the 8700k would have been quad core if Ryzen sucked.

 

[Shamrock]

3900XT全核能上4.6？有这么猛吗？ - 电脑讨论 - Chiphell - 分享与交流用户体验

3900XT全核能上4.6？有这么猛吗？,不知道是真的还是假的，感兴趣的可以去看看https://www.cpu-monkey.com/en/,电脑讨论,讨论区-技术与经验的讨论 ,Chiphell - 分享与交流用户体验

www.chiphell.com

"laughs"
Is these benchmarks are really true and the latency improved a bit, makes me feel bad for Intel.

A 3600XT matching a 3950x? :O (Yes, I know it's single core score)

 

[Hitman928]

For those who may care, here's the quotes from AMD for Epyc margins.

From the 2Q19 earnings call when asked about the margins for the new 7 nm products.

OK. Yeah. So on the margin side, the new products, as we have said previously, in aggregate are greater than 50% margin.

So the new 7nm products in aggregate are over 50% margin. They don't specifically call out Epyc, but they later give us more info when asked in the 3Q19 earnings call about which products provide the greatest leverage for increasing margins.

I think if you look at the products, definitely, the new products and the 7-nanometers are very good tailwinds for the gross margin. But also, the mix of the business comes into play. The more data center revenue we capture in terms of market share obviously helps the gross margin.

Finally, in the 1Q20 earnings call they again call out data center products (Epyc) as having significantly higher margins than their corporate average (46% that quarter).

So semi-custom ramped the back half, and that does impact the gross margin being lower than corporate average.

But data center strength, as Lisa just referenced, that we are pleased with the ramp in the data center business. In data center business, the margins are significantly higher than corporate average, and has the offset to help us deliver as we guided for 2020, the 45% gross margin for 2020.

So, with 7nm products in total having over 50% margins and that AMD says increasing data center revenue is key to increasing margins and that data center revenue is almost entirely Epyc, we can add that to the fact that data center margins were significantly above the 46% corporate average, it's pretty easy to see that Epyc gross margins are well over 50%.

 

[Rigg]

What does dram calculator read your bdie quality as ?
View attachment 21611

My 3200MHz CL14 kit.

View attachment 21622

I've got you both beat

 

[.vodka]

https://www.reddit.com/r/overclocking/comments/grm5r5

I have a "day 1" Ryzen 3600 (dated 1922). I've done a slightly daft thing and purchased another Ryzen 3600, it just arrived, and is dated 2009 (dates are YYWW in case anyone is wondering).

The previous one wouldn't boot at 4.2 GHz 1.25V, so I didn't bother - just ran it on PBO where it would happily hold +100 MHz offset and was working fine ever since.

This one can do Cinebench runs at 4.6 GHz 1.268V

Note that it's not fully stable as OCCT reboots after a minute or so at that setting, and I'm not comfortable pushing it much past that point voltage wise. But I've dropped 50Mhz to 4.55 GHz 1.268V and it's not yet done anything funny.

I've seen several posts about ridiculous 3600 overclocks so I wanted to verify it myself. It's 100% true, those chips can do crazy clocks compared to release day ones.

Edit: I eventually settled at 4500 1.275 because CB R20 was not passing on 4550 for whatever reason, and I want it stable.

Would you look at that.

Batch # 2009 (9th week, that's February-March), can do cinebench runs at 4.6GHz ~1.25v all core, 4.55GHz can at least do some stress testing. He finally settled at 4.5GHz 1.275v

It also does 1900MHz FCLK.

The improvement since launch silicon is insane. I'd say the batch currently in production must be even more capable, we should start hearing about these shortly (~July?)

 

[Kenmitch]

I've got you both beat
View attachment 21718

Nice!

What was yours sold as?

 

[JPB]

AMD's new Ryzen 9 3900XT beats Intel Core i9-10900K in this benchmark

AMD's upcoming Matisse Refresh is going to kick some Comet Lake ass when it drops

It came out of nowhere, then we began hearing non-stop news about AMD's new Matisse Refresh and now we have some early leaked benchmark results on the purported Ryzen 9 3900XT, Ryzen 7 3800XT, and Ryzen 5 3600XT processors.

The new benchmark results are coming from CPU-monkey, which has a database with a bunch of processors and their performance numbers. AMD hasn't confirmed these Matisse Refresh CPUs, and we're looking at leaked benchmarks -- so as always, prepare your salt shakers.

As for the benchmarks we're looking at Cinebench R20 results in single-core tests, where AMD's purported Ryzen 9 3900XT beats Intel's new Core i9-10900K. But more interestingly (at least to me) is the mid-range Ryzen 5 3600XT battles the higher-end Core i9-10900 and beats it, the Core i9-10900KF and beats it, and it also beats the Core i7-10700KF and Core i7-10700 processors. What a champion.

This is of course just Cinebench R20 and nothing else, no games -- but still, AMD is flexing some Zen 2 muscle right here. This is in single-core benchmarking, but AMD and its Zen 2 processors excel at multi-core tests which once again, AMD slays in early Cinebench R20 multi-core results.


Cinebench R20 multi-core results

Now remember these are leaked results and nothing is official just yet, but here's what we're looking at when it comes to Cinebench R20 multi-core tests:

AMD Ryzen 9 3900XT: 7479 points
Intel Core i9-10900K: 6399 points
AMD Ryzen 7 3800XT: 5297 points
Intel Core i7-10700K: 5292 points
AMD Ryzen 5 3600XT: 4007 points
Intel Core i5-10600K: 3647 points

AMD "Matisse Refresh" CPU Specs (So Far)

These specs could change, and I'm taking into consideration that AMD won't make any adjustments to the core count of its Matisse Refresh processors. I doubt we'll see core count changes, as the CPU frequency hike should be enough to entice consumers.

AMD Ryzen 9 3900XT: 12C/24T @ up to 4.8GHz
AMD Ryzen 9 3900X: 12C/24T @ up to 4.6GHz
AMD Ryzen 7 3800XT: 8C/12T @ up to 4.7GHz
AMD Ryzen 7 3800X: 8C/12T @ up to 4.5GHz
AMD Ryzen 5 3600XT: 6C/12T @ up to 4.7GHz
AMD Ryzen 5 3600X: 6C/12T @ up to 4.4GHz

AMD is expected to unveil its new Ryzen 9 3900XT, Ryzen 7 3800XT, and Ryzen 5 3600XT processors on June 16 with a retail release of July 7 (7/7, just like the original release of the Ryzen 3000 series CPUs)