AMD Bristol/Stoney Ridge Thread

Page 48 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
So they are literally going to wipe the floor of all the established and well running 14nm zen equipment, process, knowledge and personel to make room for building up 12fdx?

Can you explain the economic rationale for such a decision? Most cost is tied in establishing process.
 

ET

Senior member
Oct 12, 1999
521
33
91
I would have thought they'd wait for the 2nd gen VCS for a drop in upgrade. Unless they get 5CU or better I think this will be a downgrade or sidegrade.

...

Another worry about VCS (not sure about this) is the RAM; I think they were waffling back and forth between 4 and 8gb.

The Indiegogo campaign had 8GB dual channel in the spec, and I don't think they can go back on that, because it would be cheating the backers.

As for the APU, yes, it's a sidegrade, but in general I'd say it's better. An Athlon 200GE beats an A12-9800 quite soundly in many games. The A12-9800 does win in some cases, but I'd say that a 2C/3CU Zen+ APU would beat the A10-9630P for most tasks.

I find it hard to believe that AMD will offer a 2C/5CU or such for this project, nor that Atari will go for something that's not low end. Though I do think that if the project succeeds Atari will try to release a better performing version (4C/8CU).
 
  • Like
Reactions: amd6502

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
So they are literally going to wipe the floor of all the established and well running 14nm zen equipment, process, knowledge and personel to make room for building up 12fdx?
12FDX can re-use 14LPP/12LP equipment, knowledge, personnel. Other than that, AMD is moving to TSMC for the totality of their premium "Zen" roadmap.
14/12LPP is processed in NY while FDSOI is at their Dresden plant, in Germany.
However, GlobalFoundries is pushing 12FDX at Malta first. With 7LP halted indefinitely, the only track available is the 14LPP/12LP to 12FDX.

Malta and Fishkill are leading edge for FDSOI and RFSOI(CTLs-FDSOI)/(SiGe HBT FDSOI) respectively. With, Chengdu/Dresden/Singapore being high volume/mainstream.
 
Last edited:

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
The Indiegogo campaign had 8GB dual channel in the spec, and I don't think they can go back on that, because it would be cheating the backers.

As for the APU, yes, it's a sidegrade, but in general I'd say it's better. An Athlon 200GE beats an A12-9800 quite soundly in many games. The A12-9800 does win in some cases, but I'd say that a 2C/3CU Zen+ APU would beat the A10-9630P for most tasks.

I find it hard to believe that AMD will offer a 2C/5CU or such for this project, nor that Atari will go for something that's not low end. Though I do think that if the project succeeds Atari will try to release a better performing version (4C/8CU).

They describe the Zen part as "thus-far unannounced", so it might not be quite as cut down in the GPU as the 200GE.
 

ET

Senior member
Oct 12, 1999
521
33
91
They describe the Zen part as "thus-far unannounced", so it might not be quite as cut down in the GPU as the 200GE.

It might be, but I think that's more wishful thinking than anything. More likely it's simply an embedded version of a new 2C/3CU die.

As I said before, I see no indication that Atari might go for more than the minimum, and AMD has never offered a 2 core variant with more than 3 CU. So while anything is possible, it seems unlikely to me.
 
  • Like
Reactions: amd6502

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
12FDX can re-use 14LPP/12LP equipment, knowledge, personnel. Other than that, AMD is moving to TSMC for the totality of their premium "Zen" roadmap.However, GlobalFoundries is pushing 12FDX at Malta first. With 7LP halted indefinitely, the only track available is the 14LPP/12LP to 12FDX.

Malta and Fishkill are leading edge for FDSOI and RFSOI(CTLs-FDSOI)/(SiGe HBT FDSOI) respectively. With, Chengdu/Dresden/Singapore being high volume/mainstream.

Why reuse 14lpp when you can just continue? If you implement new process even a variant you take personel out to make the transition.

If you continue making some cut down zen 14nm apu you address the need of the wsa. Seems pretty straightforward to me.

I get the good argument 12fdx is the only path to improve I just don't see how the cost of the transition can be covered with lowest end products. On the other hand the wsa expires in 2024 as I recall so one can say they need to do something to use all that low end process then.
 
  • Like
Reactions: amd6502 and NTMBK

amd6502

Senior member
Apr 21, 2017
971
360
136
Yes, in the whole mobile or desktop lineup all 2C only got Vega 3, but unless there's a technical reason for this, I don't think they need to limit the VCS box to Vega 3. Product segmentation isn't really relevant here and it's kind of in AMD's interest that VCS has decent chances of succeeding, as they could become a much bigger customer for their 2nd generation.

The defect rate can't be so large that they need to be so aggressive in disabling the GPU. They might also lower the freqs maybe to make more CUs pass. (eg.: 6CU at 700-800MHz maybe and call it Vega 4) Anyway, i think a capable GPU ranks top importance for this console.

Also, we still have no sign of a native dual core, and at this point if there were, a Zen+ dual core APU makes much more sense.

As for FDX also still no sign of anything in the works or about to be released. Also, these products might be limited to GCN 1.2 (or they would need to port Vega or Navi just to have a small iGPU).
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Why reuse 14lpp when you can just continue? If you implement new process even a variant you take personel out to make the transition.
AMD has 14LPP, 14LPP+, 12LP covered. Then, there is process-agnostic for FinFETs; 14LPP/12LP Boost and 14LPP/12LP Boost Plus.
If you continue making some cut down zen 14nm apu you address the need of the wsa. Seems pretty straightforward to me.
The issue is that the harvested and smaller 14nm APU already has a successor on 7nm! Everything with Zen is moving to 7nm.

Raven2 and I/O chips isn't going to save AMD from paying GlobalFoundries. Especially, with the 7nm lineup intruding. Why buy Zen, when Zen2/Zen3 will offer better performance at same cost and smaller dies, or same performance for lower power and significantly smaller dies.

AMD can bypass the volume issue of 2020/2021 with a 12FDX lineup that targets;
>$50 for an 8-core CPU. <-- large market
<$50 for a 4C/8CU APU <-- big market
<$25 for a 2C/3CU APU <-- big market
>$25 for a 10CU GPU <-- large market

The best way to ensure volume is to sell new products at newer markets that are growing. Extreme budget, AArch64 per x86-64 in mobility is 50:1, non-first world markets are exploding for low cost.
 
Last edited:

ET

Senior member
Oct 12, 1999
521
33
91
Yes, in the whole mobile or desktop lineup all 2C only got Vega 3, but unless there's a technical reason for this, I don't think they need to limit the VCS box to Vega 3.

I don't think there's a technical reason for it, but if the 2C/3CU chip performs better than the A10-9630P in most games, and I think it would, then there's no marketing reason to go for anything higher. If backers are already getting an APU that's faster, as well as better featured and lower power, what reason does Atari have to give them anything better for the same price?

(Even if the 2C/3CU Zen chip isn't faster than the A10-9630P, my view of Atari is that it's a scummy enough company to simply tell everyone that it is. But I think that the Zen APU will beat the A10-9630P on average.)

Which is why I can't imagine a need for a 2C/>3CU variant in the current scenario. A higher end Atari model could have a higher and chip with more cores (on both the CPU and GPU sides), but since Atari already pegged the CPU cores at 2 for this one, guessing that there will be 3 CUs is reasonable, IMO.

Also, we still have no sign of a native dual core, and at this point if there were, a Zen+ dual core APU makes much more sense.

We've had signs, just no conclusive proof. And yes, I'm guessing that Atari would go for a native 2C/3CU chip if there is one.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126

I bought a Stoney Ridge Dell Inspiron 3180 Windows 10 laptop recently. It has 4GB of DDR4 (single-channel), and 32GB eMMC storage.

It's a DOG! Web-browsing is fine enough, if that's the ONLY thing that you're doing at the time. Try for a little bit of multi-tasking, and BOOM, the thing struggles like heck. Of course, that's makes sense, since the A6-9220e is a SINGLE-CORE CPU, effectively. (Single-Module, dual integer threads, single FP thread).

Throw Skype into the mix, while web browsing with Firefox, and the thing GRINDS to a halt. Since I do that often, I sometimes don't pick up that laptop, I pick up my Intel Atom N2830-based dual-core (a TRUE dual-core CPU) to use instead. It CAN handle web browsing and Skype at the same time.

The sooner that these pre-Zen architecture, GARBAGE AMD APUs are gone, the better!

Not sure why you have this sick fascination with BullDozer-derivative architectures, Nosta, but it's entirely unhealthy. Those CPU cores aren't viable for today's 64-bit multi-threaded software. They just simply aren't. The fact that AMD can crank them out for cheap due to the WSA requirements, is IMHO the only reason that this garbage is sticking around.

Edit: It should be noted that the graphics half of the APU is fine, it even supports VSR (virtual super-resolution), enabling me to view Windows 10 in 1080P, on a physically 1366x768 screen, and still be VERY readable.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Not sure why you have this sick fascination with BullDozer-derivative architectures, Nosta, but it's entirely unhealthy. Those CPU cores aren't viable for today's 64-bit multi-threaded software. They just simply aren't. The fact that AMD can crank them out for cheap due to the WSA requirements, is IMHO the only reason that this garbage is sticking around.
Bulldozer is a gateway to a bunch of variable Cluster-based Multithreading designs. The big guns being in covert mode is;
Two computational execution clusters each with 4 ALUs(GP = INT), 4 FP128 FMACs, 4 Int128 iFMACs(FP + Int = VEC)
Two memory execution clusters each with 2 LD/ST AGUs + 1 Gather Unit + 1 Scatter Unit.
(A thread can exist simultaneously on total cluster A and B.)

Improving on CMT is cheaper than making big(ger) SMT cores. There is plenty of patent fodder that AMD released between 2014-2016 for successors of Excavator. Which basically means that there is a post-Excavator CMT design.
Throw Skype into the mix, while web browsing with Firefox, and the thing GRINDS to a halt. Since I do that often, I sometimes don't pick up that laptop, I pick up my Intel Atom N2830-based dual-core (a TRUE dual-core CPU) to use instead. It CAN handle web browsing and Skype at the same time.
Have you tried disabling spectre protections? I don't see how it can feasibly be slower. Do to various benchmarks pointing towards near 2x performance in Multi and Single. It could also not be the APU, but the system configuration.

---
Closer to Vmin, FDX is faster than custom 28nm. Closer to Vmax, FDX is faster than custom 28nm. Around Vnom, FDX is faster than custom 28nm. The shrink to 22FDX and more with 12FDX, can make the gross die per wafer + wafer cost = total wafer cost provide higher margins at lower costs.
 
Last edited:

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
Bulldozer is a gateway to a bunch of variable Cluster-based Multithreading designs. The big guns being in covert mode is;
Two computational execution clusters each with 4 ALUs(GP = INT), 4 FP128 FMACs, 4 Int128 iFMACs(FP + Int = VEC)
Two memory execution clusters each with 2 LD/ST AGUs + 1 Gather Unit + 1 Scatter Unit.
(A thread can exist simultaneously on total cluster A and B.)

Improving on CMT is cheaper than making big(ger) SMT cores. There is plenty of patent fodder that AMD released between 2014-2016 for successors of Excavator. Which basically means that there is a post-Excavator CMT design.
Have you tried disabling spectre protections? I don't see how it can feasibly be slower. Do to various benchmarks pointing towards near 2x performance in Multi and Single. It could also not be the APU, but the system configuration.

---
Closer to Vmin, FDX is faster than custom 28nm. Closer to Vmax, FDX is faster than custom 28nm. Around Vnom, FDX is faster than custom 28nm. The shrink to 22FDX and more with 12FDX, can make the gross die per wafer + wafer cost = total wafer cost provide higher margins at lower costs.
Larrys experience mirrors the bm. I think its telling. Bd derived cores is just crap vs a modern core like zen. It's only on paper the BD core rocks.

Why erode your brand with such a bad user experience? This product is - at best - only suitable for cheap chromebooks. And even here it's time is limited.

The user experience between a single module excavator and a 2c zen is huge. It makes no sense to go competing with arm eco system with a unbalanced long pipeline failed server arch.

Amd needs to stop jogging around in that red ocean. Hopefully BR chome edition is the last run.
 
Last edited:

amd6502

Senior member
Apr 21, 2017
971
360
136
I bought a Stoney Ridge Dell Inspiron 3180 Windows 10 laptop recently. It has 4GB of DDR4 (single-channel), and 32GB eMMC storage.

Larry, swap that 4gb module with an 8gb or 16gb and I'll bet you it will not grind to a halt. However, yes, it will be damned slow (but on par with the "true" dual core atom). Once you're swapping to HDD or even eMMC you will grind to a halt. But yes, dual thread at sub 10W (or even 15W sometimes) and multitasking is asking for trouble if you expect anything but sluggishness. Only fast frequencies (>3.5ghz) or high IPC will offer relief from the context switching, time sharing, and low cache hits problems you see when active threads greatly outnumber available threads.

Another tip, go to the firefox preferences and make sure you have 2 threads only option for the multithread browser compute. The default setting firefox ships with seems to be 4 threads. On low RAM systems, a high number of threads is not good here, because of considerably higher memory usage. And running more threads than computationally available is almost always a bad idea (some exception when there are high latency involved, eg big compile projects with heavy access to slow old hdd).

I don't think there's a technical reason for it, but if the 2C/3CU chip performs better than the A10-9630P in most games, and I think it would, then there's no marketing reason to go for anything higher. If backers are already getting an APU that's faster, as well as better featured and lower power, what reason does Atari have to give them anything better for the same price?

If VCS is running dual channel, the getting this free easy 50%+ more GPU performance would make a world of difference in my opinion (quite a bit more important than sight diff in cpu performance like seen in BR vs athlon 200). Why limit the number of titles and settings they're played at when there's no reason not to. There is no product segmentation concern here. It's a no brainer for AMD as far as I'm concerned. The only explanation I can think of is that there might be technical reasons; like it would make the drivers or APU management much more complex.
 

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
But yes, dual thread at sub 10W (or even 15W sometimes) and multitasking is asking for trouble if you expect anything but sluggishness. Only fast frequencies (>3.5ghz) or high IPC will offer relief from the context switching, time sharing, and low cache hits problems you see when active threads greatly outnumber available threads.
Try 6W, 1.6Ghz. :(
 

amd6502

Senior member
Apr 21, 2017
971
360
136
Try 6W, 1.6Ghz. :(

Well, try making sure the threads for Firefox web content is set to 2. And use the taskmanager graph or app to keep an eye on how much free memory you have.

Windows 10 seems to do memory compression, which helps, but again this uses another thread. That's why more than 2 threads are very good these days.

I was temporarily using a low binning stoney a6-9225 but at least that ran at 15W (2.6-3ghz). It wasn't fast but quite decent and useable even with 20+ tabs. But you had to keep an eye on the RAM which, as a bottom $$ notebook, was only 4GB. That was the most limiting factor. I think it was an ideapad 330, and it came with only 1 sodimm slot that had a 4gb module. (So a bit of a pita if you want to upgrade to 8gb.)

I would pass on a 6w version unless it was a top binning Stoney. A decent binning at 10w would also be okay.

Dang Larry, 1.6ghz sounds bad. It literally is craptop material. Sorry man, maybe just limit it to easy single application at a time and stay under not much more than a dozen tabs. On the bright side, at least nobody would steal that at a coffee shop or hostel. So could be an appropriate travel laptop.
 
Last edited:

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Bd derived cores is just crap vs a modern core like zen.
Well, that is a comparison of products of largely different eras.

The general idea of a FDX product is that the density and perf/watt would catch up.
28nm 9T -> 22FDX 8T => ~0.7x Logic scaling / ~0.8x Register/SRAM scaling
^-- this is enough area to complete the ALUs in the AGUs. To get four EX units, two are complex(MUL/DIV) and two are Agen focused. This is relatively minor to what they did with the FPU between BD and SR. As the integer scheduler is already quad-ported in the cores. Arithmetic(EXs) and logic(AG Logic Us) units share a reservation station. Since, there is two cores and no sharing on the important bits, 8 EXs is equivalent to 2-cores of Zen's EX.
22FDX 8T -> 12FDX 7.5T => ~0.6x Logic scaling / ~0.55x Register/SRAM
^-- this shrink allows for the architecture to use Zen's tiles/macros/datapaths. This comes with the added advantages...

So, physical design comes from Zen which is 14LPP 9T and is ported and optimized to 12FDX 7.5T. The architecture baseline comes from 28nm and is ported to 12FDX 7.5T.

Enhanced Integer IPC. The FPU is up in the air if it will Zen-derived; 2x FMAC via bridged 2 MUL and 2 ADD or XV-derived; 2x FMACs via 2x Fused MACS. FDX has single-well and mixed-well SRAM/PRFs/CAMs, etc. The FDX design can be quite a bit more dense than the FinFET physical design.

Basically, a 12FDX CPU can be at worst a speedy and efficient Excavator-esque module and at best a module with two Zen-esque cores with a single Zen FPU. The latter is actual much closer to the original Cluster-based Multithreaded design for the 2005-slide.

An Excavator core has two macro-op decoders with two cycles to get four macro-ops.
Whereas a Zen core has four macro-op decoders which operates at same principle as the Bulldozer's module four macro-ops. Two threads get 4 macro-ops every two cycles.

Excavator-derived core rigged with Zen IP, still fits the portfolio of Zen-lite.

22FDX => 50% lower power w/ same design
12FDX => 50% lower power w/ same design
Redesign/New design => 60% lower power
Rerouted and replaced => 40% lower power
Improved/Redesigned power delivery => 30% lower power
so on, so forth.

1. The product will have the same paper specifications(core count + compute unit count) as Bristol Ridge and Stoney Ridge.
2. The product will be on a newer node, specifically 12FDX.
3. The product will have higher power efficiency targets than Zen/Zen+, however lower total performance targets than Zen/Zen+.
4. The product will cost less than its counterparts, example;
Renoir(1xx mm2 die) => Less than or equal to $160 // 12FDX BR(1xx mm2 die) => Less than or equal to $60(closer to $40-$50)
Raven2(1xx mm2 die)/Renoir2(0xx mm2 die) => Greater than or equal to $60 // 12FDX SR(0xx mm2 die) => Less than or equal to $30(closer to $15-$20)
5. Do to GlobalFoundries more aggressive ramping, the production of the lineup is no longer tied to a single foundry location. If AMD wants to target a specific region, they can target the region with the fab. Examples being Chengdu => APAC/Greater China, Malta => Americas, Dresden => EMEA.

FDX => Cost-effective performance // Cost-effective -> Economical in terms of the goods or services received for the money spent.
FinFET => Performance and density at any cost. // at all costs (or at any cost) -> In any way possible, regardless of risk or expense.
Just wanted to post this.

There was also an extremely close relationship between 12FDX and 7nm;
- Development on planar and finFET technologies: 22nm 14nm 12FDX 7nm at GF Fab 8 / Fab 10
- Pre-clean chamber and Producer and Precision CVD products for customer 7nm and 12FDX technology. <-(customer is GlobalFoundries)
It is also closely related to the FinFET nodes overall;
- 14LPP/12LP/12FDX Device Director
- Working on 7nm,14LPP,12FDX
- •GLOBALFOUNDRIES 14LPP Mobility Solution ... •GLOBALFOUNDRIES 7nm FinFET Mobility Solution ... •GLOBALFOUNDRIES 12FDX SOI Technology
- Experienced Tech node: 7nm, 12FDX, 14nm.

12FDX European project is known,
12FDX American project is known,
12FDX Chinese project is I don't know; GloFo's SVP of Asia business development, Americo Lemos => Looking ahead, Lemos said that GloFo expects to roll out its 12nm FD-SOI (12FDX) process as early as 2019. => 19 February 2019, 12:11, DigiTimes
All three should hopefully be producing 12FDX by the end of 2019.

- If late last year was v0.2 qualification PDK for 12FDX
and Q1 2016 was v0.2 for 22FDX, Q4 2016 for v1.0
Then, v1.0 should be Q3/2H 2019.
 
Last edited:
  • Like
Reactions: amd6502

amd6502

Senior member
Apr 21, 2017
971
360
136
So lower project cost is a pro for FDX. But having two architectures to worry about when there are intense challenges of fixing Spectre class vulnerabilities probably figures that they concentrate 95% plus of design team on Zen2 and on.

Personally, I think XV is refined to a high extent already, as a low transistor count core it is just about perfect already; BR could be super suited in quadcore 2+2 big little like configuration even with no physical changes. They could make just minor changes and merge the cache system and GPU of stoney with the 4 thread ability of BR and it would be an atom killer on the low end spectrum. 4 threads on 10W with medium binnning 22FDX or high binning 28nm bulk.

Alternatively, do what krumme and ET suggest and bring out a low power mobile oriented 10W and sub 10W dual core zen+ APU (that secondarily also will handle higher wattages like 25-30w for AM4 packaged athlon 300 ge's). And picasso already has many efficiency gains.

The mobile APU is by far the largest mass market for 2c/4t. So that option of getting a Picasso-L is pretty likely.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
I somehow missed the Lenovo versions.

https://www.lenovo.com/us/en/laptops/lenovo/student-laptops/Lenovo-14w/p/88EL10S9990
https://www.lenovo.com/us/en/laptops/lenovo/student-chromebooks/Lenovo-14e-Chromebook/p/88ELC1S9991

--old
Stoney Ridge Chrome;
- https://store.hp.com/us/en/pdp/hp-chromebook-14-db0020nr
- https://store.hp.com/us/en/pdp/hp-chromebook-14-db0030nr
- HP Chromebook 11A G6 6KJ20UT#ABA
- https://www.acer.com/ac/en/US/content/professional-model/NX.H8SAA.001
^-- 1366x768 and A4-9120C(equiv A6-9220e)

- HP Chromebook 14A G5
- https://www.acer.com/ac/en/US/content/professional-model/NX.H8SAA.002
^- 1920x1080(HP it is an option) and A6-9220C(equiv A9-9420e)
old--

There is apparently a couple more Acer models slated for Q2 2019;
Acer Chromebook™ 311 -> A4-9120C
Acer Chromebook™ Spin 311 -> A4-9120C
Acer Travelmate B1 -> A6-9220C

Both the Acer Travelmate B1 and Lenevo 14w have options for a m.2 256GB dual-lane(2x) PCIe NVMe SSD and are Windows/1080p(Acer its an option).

Just a thought, I wonder how much savings the back to school sales will be. Especially, since these are aimed at the Education field.
 

amd6502

Senior member
Apr 21, 2017
971
360
136
Well, it's good that they reversed the A-number inflation back from A6-A9 to a more appropriate A4-6, but it would be better if they'd avoid the 6W A4 altogether and have it be a 10W part. Now if OEMs put A4 strictly in tablets only then it wouldn't be so bad, but you can't expect OEMs to behave. They are going to screw some consumers and AMD will pay a portion of the blowback with the reputation on the A-series.

Not impressed with Lenovo here. Any OEM that will solder on a piddly 4GB without offering an expansion sodimm slot does not score any positive points with me.

I assume HP + acer is going the same route on chromebooks. Built in obsolescence probably among top priority feauters for this market segment.

I also am guessing there is not sata bay or an alternative secondary internal storage method.

As somebody interested only in one step above the tablets and disposables segment I hope we get a 4 thread 10W and below oriented design soon.
 
Last edited:

amd6502

Senior member
Apr 21, 2017
971
360
136
I imagine that the 200GE is purely binned first gen Raven Ridge. It was said that Raven 2 has 4x PCIe, and, assuming that's true, it would conflict with 200GE's specs. I figure we'll have to wait for the 300GE to see Raven 2.

Not sure what is meant by 4x pcie. (only four pcie lanes? I suppose that would be okay for a budget notebook/chromebook).

However, on the topic of Raven2....

I just built an new kernel (as I'm moving up to linux 5!) for my laptop today and noticed under the missing gpu firmware warnings "raven2" was listed (among all other generations of APUs and dGPUs). Now this could simply be the 3CU versions seen from RR salvaged 3CU APUs (as techpowerup seems to have dug up: https://www.techpowerup.com/gpu-specs/amd-raven-2.g888 ). Or it might be another project. Picasso was also already listed as well. Too bad nothing conclusive can be drawn from this.
 

ET

Senior member
Oct 12, 1999
521
33
91
Yeah, PCIe 4x means it only has 4 PCIe lanes, compared to 8 for Raven Ridge. That was the spec rumour, and it makes sense for a cut down die which isn't likely to be put alongside a discrete GPU.

Raven 2 and Picasso have been in the Linux code for about half a year, as reported by Phoronix. When Ryzen Mobile 3000 was released Ian Cutress asked AMD about the Ryzen 3 3200U and Athlon 300U being 14 nm and IIRC the answer implied that it was a new chip distinct from Picasso.

That's as far as rumours regarding a cut down RR exist (far as I know). I haven't seen any 3200U or 300U based laptop announced, so it's impossible to test if this is indeed a new die or just salvaged RR. The only significant difference in the specs is that the 3200U is certified for 105°C operation (like the Picasso chips), while the 2200U only for 95°C (like RR), but I'm not sure this says much, and could very well be just copy and paste and not a real spec.
 
  • Like
Reactions: amd6502