AMD Bristol/Stoney Ridge Thread

Page 51 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

amd6502

Senior member
Apr 21, 2017
971
360
136
I think it took almost 200W for Piledriver to hit 5GHz (well, maybe 100W for the boosting core and the rest for the uncore and other 7 cores). Even with the advancements, I don't think this is doable at ultra low power.

Following the news today on the new R1000 (and V1000 series), it's starting to look like there will be an upcoming native dual core APU die (either now, or about 6 months down the road), based on Picasso. This should be Zen+ ~100mm2 group of products, optimized to address the ~10W market (from 6W to 15W).

At 2c/4t it should be a little more capable than BR at CPU tasks, but have only well under half the GPU perf of BR. Should have almost 3B transistors (vs 1.2B for Stoney and 3.1B for BR) and so it should still cost significantly more to produce than Stoney yet not too significantly more than BR (which is why Stoney remains in production while BR probably does not).

Lower production costs to the point of recovering development costs and product premiums would be needed to justify an FDX alternative positioned below Picasso-L. I think a 4 thread Stoney++ on 22FDX could still possibly do that, but the odds of seeing this in the future appear to be low.
 
Last edited:
  • Like
Reactions: NTMBK

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
I think it took almost 200W for Piledriver to hit 5GHz (well, maybe 100W for the boosting core and the rest for the uncore and other 7 cores). Even with the advancements, I don't think this is doable at ultra low power.
The voltage scales across the SOI nodes. It is also easier for FDSOI to get really low delay and thus really high clocks compared to PDSOI/FinFET.

32nm PDSOI <1.6V for 5 GHz -> 28nm HP-FDSOI <1.4V for 5 GHz -> 22nm FDX <1.2V for 5 GHz -> 12nm FDX <1.0V for 5 GHz. The FDSOI nodes have FBB which can be used to increase speeds at lower voltages. 28nm HP-FDSOI <1.3V for <5 GHz w/ FBB -> 22nm FDX <1.1V for <5 GHz w/ Conventional FBB -> 12nm FDX <0.85V for <5 GHz w/ DITO FBB.

Scaling wise, with increased energy efficiency pulled across from Zen's logic and memory compilers, etc. What is 220Ws in 32nm PDSOI with energy optimization would be 15W~35W in 12FDX. Absolutes is 22FDX on power is 1/5th 32PD, and 12FDX in turn can be 1/2th of that. With all the optimizations on bulk(SR/XV) and some agnostic(Zen). Eight core with no UMC(universal memory controller; MCT/DCT/PHY) and only some instances of infinity links. Easy, efficient, and cheap alternative to Zen, in CPU chiplet cases. While, pushing SKUs in likeness of Bristol/Stoney into extremely efficient budget parts.

Under the same math, since 28nm bulk on the low power is similar to PDSOI. So, a9-9420e(9220c)/a6-9220e(9120c) is 1.0V is 2.7/2.4 GHz, 0.9~1.0V is 2.0 GHz. Which would place a 12FDX product at 0.4V for 2 GHz and with FBB w/o power increases to ~2.5 GHz. At that voltage a 6W Stoney Ridge should scale to 0.6W~1.2W. This product does however need some stuff from Zen;
- Excavator's stock voltage range is 0.7V to 1.35V. This needs to adopt Zen's 0.4V to 1.0V stock voltage range.
- Excavator has a less accurate AVFS system and it might be slower at switching than Zen's. This needs to not only adopt it but modify it as well for faster freq/volt/bias switching and to add ABS for AVFBS. Both of which are FDSOI exclusive, body-bias and the shorter time needed for going between power states, ex: Px to Py.
 
Last edited:
  • Like
Reactions: amd6502

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Not sure if it matters but the cost to buy/bid for all the new stuff is near these values;
Ryzen 3200U is $100
Athlon 300U is $80
A6-9220C is $30
A4-9120C is $25

//Misc 12FDX stuff;
Lib cells from 28nm to 12FDX should be around 0.38x smaller?, still pending.
6T SRAMs are on-par with Intel; 14LPP/12LP SRAM is 0.0806 mm2 to 12FDX/Intel's 14nm is ~0.058mm2 which is a 0.72x shrink?
 
Last edited:
  • Like
Reactions: amd6502

amd6502

Senior member
Apr 21, 2017
971
360
136
Not sure if it matters but the cost to buy/bid for all the new stuff is near these values;
Ryzen 3200U is $100
Athlon 300U is $80
A6-9220C is $30
A4-9120C is $25

//Misc 12FDX stuff;
Lib cells from 28nm to 12FDX should be around 0.38x smaller?, still pending.
6T SRAMs are on-par with Intel; 14LPP/12LP SRAM is 0.0806 mm2 to 12FDX/Intel's 14nm is ~0.058mm2 which is a 0.72x shrink?

Seems like too small a premium between Stoney top bin (A6) and medium (A4) bin; the boost difference is 12%, 300Mhz, 2.4 vs 2.7ghz.

That's a good cost advantage. So Stoney around $30. I'd guess Picasso-L around $50-$60.

Compared to Pentirum they are not doing too bad at the moment, just because Stoney can go down in price so much (being just ~100mm2 at what is now a very old node) while Pentirums right now have a shortage of 14nm capacity. The main deficit is really the number of threads:

http://www.cpu-world.com/Compare_CP...,Intel_FH8066802979803,Intel_FH8066802980002/

I think pentirum may also have low load and idle efficiency advantages, just because the atom cores can clock down so low (somewhat below 600mhz if I remember right from using something like an n2820).

Both number of threads and low p-state wattage is something that I think a next gen jaguar with on-demand coarse grain multi-threading could help with. Probably it won't happen, and the low end x86 mobile market will ride the dual thread as long as it's still acceptable.

It would make sense if Stoney would in fact get company from 6W single core Zen (1c/2t) Picasso-L die salvage sometime in 2020; then Stoney production would end within half a year to a year of that, and all dual thread x86 consumer SoC would cease to be produced early 2022 and fade into history that year.

The other approach, of which there are some hints, is to revive Puma (maybe on FDX). At 28nm it was about 4 cores in under 15mm2. So octacore in 30mm2. Salvaged quads can be funneled into 10W and below market, while hexacores and 8c go to the other markets. This was some leak by tum_apisak who has sprung proper leaks before. Search "A9-9820" and there is a reddit thread + various articles like https://www.guru3d.com/news-story/a...820-reveal-themselves-in-3dmark-database.html

So RX models go to consoles, while A9 models go to things like all-in-ones and even laptops, and maybe even two or three AM4 packaged skus. The GPU might be comparable to almost to an R7 250, maxing out DDR4 2133 or 2400 dual channel, being something like 5 or 6 CU at 1300MHz. The really bad single thread and small die area of ~200mm2 at 28nm (or less at 22nm) would make this family strictly all budget.
 
Last edited:

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
There are limiting factors for 22nm FDSOI for next-gen CMT;
-> Not a big enough shrink.
-> Doesn't provide the performance/power of 7nm.
-> Would be a hard sell to sell with 7nm APUs.

12FDX would have a big enough shrink, provides the performance/power of 7nm. Making it an easier sell at the lower cost of FDSOI on the same node generation. When the 7nm (CPUs+)APUs come out the 12FDX (CPUs+)APUs would be a cost-effective and viable alternative. As they would have the same generation worth of power efficiency.

Jaguar/Puma is on 14nm, and it is more likely to be ported to 7nm because of Zen.

Zen/Jaguar is already on 14nm/7nm, which the cost of which requires extra expenditures to make a move to 22FDX/12FDX. However, Excavator had only a canned project to 20LPM. Making a next-gen to 12FDX that in fact can re-use that IP plus some of the added IP from Zen/Jag of 14nm. Which means a move to 12FDX re-utilizes depreciated expenses from past projects. The most viable candidate for a FDSOI project is one that does not have a FinFET alternative. Which of course is Carrizo/Bristol/Stoney as there is no shrink or next-gen version of the CPU core used.

Jaguar is stuck holding Zen's hand, while a CMT core is not. Imagine a CCX with Jaguar w/ no shared L2. 2x 32KB L1i/d, 1x256 KB L2, and whatever the behavior of independent 512 KB L3s are. Four cores of Zen or Eight cores of Jaguar, similar L3 CCX, different at L2 and below only. Same node, with same CCX characteristics optimized for Zen and Jaguar is the best outcome. However, a CMT design has a bigger L2 and probably would use a different complex than Zen/Jaguar.
---
RX/A9 Cato's are 14nm?, Xbox-derived but not on TSMC? My assumption is that it is its own die with 8-core + 8 CU + 128-bit DDR4. Hence, the 9820 model number, and checking the Xbox One S numbers. It should be in between 35W FX-9830P and 65W A12-9800 for its TDP.
 
Last edited:

amd6502

Senior member
Apr 21, 2017
971
360
136
There are limiting factors for 22nm FDSOI for next-gen CMT;
-> Not a big enough shrink.
-> Doesn't provide the performance/power of 7nm.
-> Would be a hard sell to sell with 7nm APUs.

12FDX would have a big enough shrink, provides the performance/power of 7nm. Making it an easier sell at the lower cost of FDSOI on the same node generation. When the 7nm (CPUs+)APUs come out the 12FDX (CPUs+) APUs would be a cost-effective and viable alternative. As they would have the same generation worth of power efficiency.

Jaguar/Puma is on 14nm, and it is more likely to be ported to 7nm because of Zen.

Zen/Jaguar is already on 14nm/7nm, which the cost of which requires extra expenditures to make a move to 22FDX/12FDX. However, Excavator had only a canned project to 20LPM. Making a next-gen to 12FDX that in fact can re-use that IP plus some of the added IP from Zen/Jag of 14nm. Which means a move to 12FDX re-utilizes depreciated expenses from past projects. The most viable candidate for a FDSOI project is one that does not have a FinFET alternative. Which of course is Carrizo/Bristol/Stoney as there is no shrink or next-gen version of the CPU core used.

Jaguar is stuck holding Zen's hand, while a CMT core is not. Imagine a CCX with Jaguar w/ no shared L2. 2x 32KB L1i/d, 1x256 KB L2, and whatever the behavior of independent 512 KB L3s are. Four cores of Zen or Eight cores of Jaguar, similar L3 CCX, different at L2 and below only. Same node, with same CCX characteristics optimized for Zen and Jaguar is the best outcome. However, a CMT design has a bigger L2 and probably would use a different complex than Zen/Jaguar.
---
RX/A9 Cato's are 14nm?, Xbox-derived but not on TSMC? My assumption is that it is its own die with 8-core + 8 CU + 128-bit DDR4. Hence, the 9820 model number, and checking the Xbox One S numbers. It should be in between 35W FX-9830P and 65W A12-9800 for its TDP.

7nm APUs will compete in totally different market (high end).

If these are xb1s salvage. You're probably right, 8CU at ~0.8ghz.

I think xb1s was using around a 30W TDP or somewhat under a quarter of xb1x. So i'd say the A9-9820 might have a 20w-40w cTDP, allowing it in mobile or SFF desktop.

Would be nice if it were destined for fm2+ and developing world (maybe in bga/fp3 form).

Jaguar is supposedly easy to port. So, if they are a new design, i think they would clock fewer CU very fast, which is something 22FDX might allow them to do withn fw body biasing. That way you end up with a smaller 6CU sub 200m2 budget die, and maybe it will still allow you to be under the max area limits that fdx requires.

12fdx project might be too costly to justify.

do 12fdx or 22fdx have something similar to the high density libraries they used for the gpu part of their 28nm APUs?

what is the cost difference between a 22fdx project versus a similar 12fdx project?
 
Last edited:

amd6502

Senior member
Apr 21, 2017
971
360
136
The 1s has just recently ditched the DVD unfortunately. Probably to push the low end gaming consoles which should appoach $200. In some ways it's competing with VCS a little now. Nintendo is in trouble.


This means it's pretty high volume production, and ongoing.

Such die salvage could lower the cost of consoles a few $ lower for msft. So I think this kind of makes sense.
 
Last edited:

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Consider it speculation;

Timeline equivalent - actual node/launch
Conesus (28nm for Chromebooks/Notebooks)/2019 -> Geneva (22FDX for Chromebooks/Notebooks)/2020 -> Ontario (12FDX for Chromebooks/Notebooks)/2021 -> Kabini (7FDX for Chromebooks/Notebooks)/2023
Stoney(FP4/FT4) -> Stoney Migration(FP4/FT4) -> New Core/SoC -> New Core/SoC

It also follows the timeline from January 2019.

The new core has multiple(two) mid_core blocks, which I have implied to mean CMT.

Migratory Stoney => 22FDX+, NMOS tSi+tensile DITO and PMOS cSiGe+compressive DITO. No body biasing impacting performance/power is expected.

GF28A Carrizo -> GF28A+ Bristol = Same thing as Stoney 28A+ to ____ 22FD+.
I recommend for further explaination => http://www.in2fab.com/22fdx.html // Assume Full Migration or Migration + Optimize. With most of the development cycle happening at ODCs before going mainline.
 
Last edited:
  • Like
Reactions: amd6502

amd6502

Senior member
Apr 21, 2017
971
360
136
Fingers crossed for that 22FDX Stoney++. Will it still be dual core (rather than quad)? If so, a third coarse grain little thread that sleeps in the L1i would greatly help it deal with multitasking. Less context switching more efficiency, and it would be a freebie with almost no transistors added.

Why no body biasing?
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Why no body biasing?
It starts off as cost...
The options are ~100 mm squared die with full redesign(w/ BB) or ~125 mm squared die with migration(w/o BB at worst).

The product also has particular needs...
1.4x CPU speed
1.3x GPU speed
1.5x memory speed
at same cost.

Bare Minimum:
Base - 1.8 GHz/1.6 GHz -> 2.52 GHz/2.24 GHz
/&/
Boost - 2.7 GHz/2.4 GHz -> 3.78 GHz/3.36 GHz
/&/
GPU - 720 MHz/600 MHz -> 936 MHz/780 MHz
/&/
DDR - 2400 MHz -> 3600 MHz (Internal I/O(NB speed) also needs to increase by 1.5x; est. 25.6 GB/s to est. 38.4 GB/s)
/&/
Cost - The same, but allows for customers to enhance their products with ease. (Same motherboards, different process)

28BLK -> 22FDX(+) would have to follow the same rules as 14LPP -> 12LP. Except, it is on high-end performance stimulants.
14LPP -> 12LP pretty much the same design with 6% increase in frequency + 10% decrease in power. <<FinFET to FinFET>>. Same BEOL/Different FEOL.
28BLK -> 22FDX the same way is a lot more performance and a lot less power. <<Planar to Planar>>. Same BEOL/Different FEOL.

Some reasons we haven't seen it sooner;
First in priority is Chengdu;
- Diffusing in China and Packaging in China is going to make it cost less.
- Phase 2 is October-December 2019 w/ both Phase1/Phase2 ramping to 85K 22FD(+) wafer starts.

Second in priority is 22FDX;
- 22FDX itself wasn't as refined as 28nm. This is observed with lower performance, higher power, increased variation, higher defect density.
- AMD was specifically waiting for 22FDX+ with its usage of 12FDX transistors and wafer. (22FDX and 12FDX do not use the same wafer, 22FDX+ and 12FDX however does)
- AMD was also waiting for it to become popular, so they don't carry the node alone.

Third in priority is Stoney Ridge and what is available;
- DDR4 has only just begun to get to 3200 MHz and above in JEDEC(1.2v).
- AMD needed an existing market and a product that can cost less. 22FDX was most expensive in 2016/2017, it had a price drop in 2018, with Chengdu ramping it might drop once again. Then in 2020, when Dresden peaks out for 22FD wafer starts for another potential price drop.
- SOITEC is another reason, FDSOI wafers are depreciating in price(-7% per year; reduced base cost = reduced overall cost). Which can be carried over to the fabless customer.

//Late edit: Just wanted this here for fun. Puma+ vs Excavator, btw.
 
Last edited:
  • Like
Reactions: amd6502 and NTMBK

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
Unless you're building on FM2+ with matching components on hand such as memory. The Athlon 200 series are limited strictly to AM4 DDR4 builds, and its GPU is also quite a bit less suited for gaming than the A8's. So the 7680 is priced decently, allowing you to bypass a low end $60+ dGPU.

The MT scaling for multiple threads of pure (double precision) floats code would be equivalent to SMT scaling. For mixed loads or pure INTs however, the scaling of CMT is very high (especially in post Piledriver with the doubled up front end).

The trouble with synthetic benches is that they rarely (if ever) give mixed loads, which are very common in real life. They will test some category, first in single thread, then in multithread. Stuff like that rarely happens in real life.

Today i had a chance to test a 7680, its awfull, with DDR3-1866 is just... bad, the 200GE smash it to bits in gaming even with just half iGPU. Yes it can play Fifa19, CS:GO (barely, 60 fps is not enoght), Fornite and Apex at 720P, but the 200GE can too, and better, specially CS:GO. Also tried the 7480, is just terrible, 100% in every game ive tested.
And i think Windows 10 is playing a card here, it should not detect Carrizo as full cores, petty sure the 9500 is better than the 7480, and the 9600 was giving a fight to 200GE due to DDR4, DDR3-1866 is not enoght for 6 CU.

At any rate, BD cores with CMT are dead, as they should, 200GE/SMT/Vega 3 wins.
 

amd6502

Senior member
Apr 21, 2017
971
360
136
Today i had a chance to test a 7680, its awfull, with DDR3-1866 is just... bad, the 200GE smash it to bits in gaming even with just half iGPU. Yes it can play Fifa19, CS:GO (barely, 60 fps is not enoght), Fornite and Apex at 720P, but the 200GE can too, and better, specially CS:GO. Also tried the 7480, is just terrible, 100% in every game ive tested.
And i think Windows 10 is playing a card here, it should not detect Carrizo as full cores, petty sure the 9500 is better than the 7480, and the 9600 was giving a fight to 200GE due to DDR4, DDR3-1866 is not enoght for 6 CU.

At any rate, BD cores with CMT are dead, as they should, 200GE/SMT/Vega 3 wins.

The GPU should blast the Vega 3. You have 384 SPs running at 1030MHz, which is equivalent to over 548 SPs at Kaveri stock clock of 720MHz. Even Kaveri A10 with 512SPs at 720Mhz needed dual channel DDR3 2133 to almost max out its iGPU. So the A8 really would do best with bandwidth provided by dual channel running somewhere between 2133 and 2400. You should double check that your configuration is dual channel, and that your drivers are also set up correcty. The A8 is the fm2+ equivalent of the A10-9700; and here is the side comparison vs the original Kaveri part versus the newer A8/A10: http://www.cpu-world.com/Compare_CPUs/AMD_AD785KXBI44JA,AMD_AD9700AGM44AB/

I also got my A8 booting on both a biostar A70M board, and a Gigabyte A68 board. So one thing I can tell you right now is that the stock heatsink on this is so small and lightweight, that it's really only suited for the 45W cTDP. At 65W setting you will hit thermal throttling very very fast; especially in non-aircondition summer environments. Put a 95W cooler (or better) on it, and this A8 should do very well.

To summarize, the stock cooler on the A8 is not gaming suitable. It is suitable really for home/office use only; or I suppose non-demanding casual gaming. For more serious gaming use, you really need to have your own proper cooling solution (old AM3 or FM1 cooler work quite well), or invest in it.

The stock cooler consists of a 70mm fan, and an square aluminum radiator of slighty larger footprint and only about 2cm (or slightly less) thickness. I believe it reduces the cost of shipping and distribution. It fits the needs of most people alright (exception maybe for non-AC&hot climates), but for gamers.... they better have their own cooler. As for the 200GE, I imagine it has the very same low end cooler, as its TDP is 35W. The 2c/4t+Vega 3 is a very efficient beast; kudos to the Zen team.

Some observations of mine: the Biostar board bios settings did not allow me to select a cTDP, so it ran at 65W cTDP only; the automatic core voltage also ran what seemed to be dangerously high, at ~1.45V and maybe higher even, while in the bios setup mode. The gigabyte did allow me to select a 45W tdp. So I'm using the gigabyte board for a SFF build; and sticking with the crappy stock fan, since it is extremely low profile, and because it won't be a gaming build.
 
Last edited:

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,691
136
And i think Windows 10 is playing a card here, it should not detect Carrizo as full cores, petty sure the 9500 is better than the 7480, and the 9600 was giving a fight to 200GE due to DDR4, DDR3-1866 is not enoght for 6 CU.

From what I've seen, Carrizo/BR's memory controller (both DDR3 and 4) is not too good either. Raven Ridges is -much- better, so that could impact things too.

In fact, Guru3D.com managed to get lowly 220/240GEs to run with memory set at 3333MHz. Quite good for such an inexpensive CPU.

https://www.guru3d.com/articles_pages/amd_athlon_220ge_and_240ge_review,18.html
 

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
The GPU should blast the Vega 3. You have 384 SPs running at 1030MHz, which is equivalent to over 548 SPs at Kaveri stock clock of 720MHz. Even Kaveri A10 with 512SPs at 720Mhz needed dual channel DDR3 2133 to almost max out its iGPU. So the A8 really would do best with bandwidth provided by dual channel running somewhere between 2133 and 2400. You should double check that your configuration is dual channel, and that your drivers are also set up correcty. The A8 is the fm2+ equivalent of the A10-9700; and here is the side comparison vs the original Kaveri part versus the newer A8/A10: http://www.cpu-world.com/Compare_CPUs/AMD_AD785KXBI44JA,AMD_AD9700AGM44AB/

I also got my A8 booting on both a biostar A70M board, and a Gigabyte A68 board. So one thing I can tell you right now is that the stock heatsink on this is so small and lightweight, that it's really only suited for the 45W cTDP. At 65W setting you will hit thermal throttling very very fast; especially in non-aircondition summer environments. Put a 95W cooler (or better) on it, and this A8 should do very well.

To summarize, the stock cooler on the A8 is not gaming suitable. It is suitable really for home/office use only; or I suppose non-demanding casual gaming. For more serious gaming use, you really need to have your own proper cooling solution (old AM3 or FM1 cooler work quite well), or invest in it.

The stock cooler consists of a 70mm fan, and an square aluminum radiator of slighty larger footprint and only about 2cm (or slightly less) thickness. I believe it reduces the cost of shipping and distribution. It fits the needs of most people alright (exception maybe for non-AC&hot climates), but for gamers.... they better have their own cooler. As for the 200GE, I imagine it has the very same low end cooler, as its TDP is 35W. The 2c/4t+Vega 3 is a very efficient beast; kudos to the Zen team.

Some observations of mine: the Biostar board bios settings did not allow me to select a cTDP, so it ran at 65W cTDP only; the automatic core voltage also ran what seemed to be dangerously high, at ~1.45V and maybe higher even, while in the bios setup mode. The gigabyte did allow me to select a 45W tdp. So I'm using the gigabyte board for a SFF build; and sticking with the crappy stock fan, since it is extremely low profile, and because it won't be a gaming build.

Its CPU/Mem issue. On Fortnite for example the 200GE and the 7680 performs almost the same BUT some textures takes a really long time to load on the 7680, meaning when you just start everything looks like crap for 2-3 mins.
Fifa 19 is CPU bound, its 100% all the time on the 7680, this causes the 7680 can run it at a little bit better FPS in Low than the 200GE can do on medium. On 200GE CPU use tends to be at aorund 50%.
GTA V is just better on 200GE at the same quality settings. Not by much but its better.
Apex about the same on both.
CS:GO was huge, it can run at almost twice the FPS on HIGH on the 200GE than the 7680 can do on low.
Dirt Rally also better on Medium@720p on the 200GE
The only instance that the 7680 was better than the 200GE was Witcher 3... but pointless since its unplayable on both.

This is really not that much of a suprise to me since the A8-9600 can beat the 200GE in gaming just by a hair. And it losses in a few, like CS:GO. I expected the 7680 to perform close to the 9600 but its not. Most likely due to slower ram and Windows seeing it as a 4C/4T CPU, it should see it as 2C/4T just like every other BD based CPU/APU... Im going to forward a bug report to Microsoft about this, just dont expect any fixes.
 
  • Like
Reactions: amd6502

amd6502

Senior member
Apr 21, 2017
971
360
136
Wow, that's interesting; thanks for the bug report. That could be helpful. I would think it'd perform like its sister chip the A10-9700.

You might try running geekbench to see if you fall in-line the norm, especially looking at the memory section of benchmarks. What brand motherbird are you using?

TechEpiphany tested this, and from the looks of it, pretty decent performance; probably roughly on par with A10-9700. He may have been using 2400 or 2133 memory.

 
Last edited:

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
Wow, that's interesting; thanks for the bug report. That could be helpful. I would think it'd perform like its sister chip the A10-9700.

You might try running geekbench to see if you fall in-line the norm, especially looking at the memory section of benchmarks. What brand motherbird are you using?

TechEpiphany tested this, and from the looks of it, pretty decent performance; probably roughly on par with A10-9700. He may have been using 2400 or 2133 memory.


He is using faster ram, at any rate he is having the same problems i noticed, very high CPU usage across the board and the GPU has problems to stay at >90% I dont think any of that is faster that a 200GE.

But acording to Cinebench the CPU is really fast for a BD...
pPsCmGa.jpg



That is almost 200GE numbers there.
JkYNTQx.png


And lots more than the 9600.
l74dczp.png
 

amd6502

Senior member
Apr 21, 2017
971
360
136
I would say that some of those games run amazingly for a properly set up A8 budget box (in 720p low to med settings); it could be great choice for someone that has 2400 DDR3 laying around. Others do tax the CPU and struggle.

In those cases, either the games are super pure FPU threads, or single+dual thread, or just optimized (-O2 or O3 compiled for core series processors rather than general x64 optimization). For cinebench benchmark 3.3x scaling is actually really very good for something that is mostly parallel very FPU heavy code. Mixed code, or pure Ints is where XV/dozers usually do their best.

The Athlon would be better in general; performance wise, and the fact that there's an upgrade path; esp for gamers. But for someone who has old parts laying around, or is building a home+office machine, I think the A8 is quite a good choice.
 

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
The games and Windows belive they have a full 4C (that includes full FPU) to use and thats just wrong, thats what happening there. The CPU itselft is kinda fast. Actually not sure why its faster than BR, it may be just a strange case on Cinebench, the 7480 was also fast compared to other duals.

The deal breaker here is ram... today is hard to find anything faster than DDR3-1600, and is just not enoght.
 

amd6502

Senior member
Apr 21, 2017
971
360
136
I'm thinking it might be something with the memory configuration (double check bios settings) maybe. I don't think it is windows managing the cores incorrectly. They essentially act like physical cores, because they pretty much mostly are. The pair of 128-bit FP units in a module however can gang up for 256-bit operations, so you will see SMT-like scaling for a pair of threads that have pure FPU code.

Even with 1600 DDR3 dual channel, it should be decent performance, maybe 20% weaker than stock a A10-7850k iGPU using 2133 RAM. So I would run geekbench and look at the perf of the memory tests.
 
Last edited:

amd6502

Senior member
Apr 21, 2017
971
360
136
I'm thinking with the start of 7nm that 28nm is on the way out this year. May be out of production in 2020. Both BR and SR with likely no SR++, at least on 28 pdsoi.
 

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
The GPU should blast the Vega 3. You have 384 SPs running at 1030MHz, which is equivalent to over 548 SPs at Kaveri stock clock of 720MHz. Even Kaveri A10 with 512SPs at 720Mhz needed dual channel DDR3 2133 to almost max out its iGPU. So the A8 really would do best with bandwidth provided by dual channel running somewhere between 2133 and 2400. You should double check that your configuration is dual channel, and that your drivers are also set up correcty. The A8 is the fm2+ equivalent of the A10-9700; and here is the side comparison vs the original Kaveri part versus the newer A8/A10: http://www.cpu-world.com/Compare_CPUs/AMD_AD785KXBI44JA,AMD_AD9700AGM44AB/

I also got my A8 booting on both a biostar A70M board, and a Gigabyte A68 board. So one thing I can tell you right now is that the stock heatsink on this is so small and lightweight, that it's really only suited for the 45W cTDP. At 65W setting you will hit thermal throttling very very fast; especially in non-aircondition summer environments. Put a 95W cooler (or better) on it, and this A8 should do very well.

To summarize, the stock cooler on the A8 is not gaming suitable. It is suitable really for home/office use only; or I suppose non-demanding casual gaming. For more serious gaming use, you really need to have your own proper cooling solution (old AM3 or FM1 cooler work quite well), or invest in it.

The stock cooler consists of a 70mm fan, and an square aluminum radiator of slighty larger footprint and only about 2cm (or slightly less) thickness. I believe it reduces the cost of shipping and distribution. It fits the needs of most people alright (exception maybe for non-AC&hot climates), but for gamers.... they better have their own cooler. As for the 200GE, I imagine it has the very same low end cooler, as its TDP is 35W. The 2c/4t+Vega 3 is a very efficient beast; kudos to the Zen team.

Some observations of mine: the Biostar board bios settings did not allow me to select a cTDP, so it ran at 65W cTDP only; the automatic core voltage also ran what seemed to be dangerously high, at ~1.45V and maybe higher even, while in the bios setup mode. The gigabyte did allow me to select a 45W tdp. So I'm using the gigabyte board for a SFF build; and sticking with the crappy stock fan, since it is extremely low profile, and because it won't be a gaming build.

If you have to pay for a big CPU cooler to go with this, it's even worse value. Much better to buy into AM4, get a stronger CPU in the Athlon, and add a cheap GPU in the future.
 
  • Like
Reactions: amd6502

amd6502

Senior member
Apr 21, 2017
971
360
136
Many upgrading from a previous fm2 or am3 build will have coolers left over that are little better than this. It has very low thermal mass. It's possible it will dissipate 65W if your room temp is 70F or lower, and you case cooling is very good. I haven't tested what sort of thermal throttling hit it might take during gaming. It probably depends on the environment quite a bit. If you are near the equator without good AC likely these are terrible. I also don't like these 70mm fans spinning up to like 5000 rpm or whatever, even if they do dissipate the 65W. I imagine it's the same cooler as in A10-9700 and Athlon 200
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
But acording to Cinebench the CPU is really fast for a BD...
pPsCmGa.jpg



That is almost 200GE numbers there.
JkYNTQx.png


And lots more than the 9600.
l74dczp.png

The 9600 should score better than this, something like 650 pts, almost as much as the 657 pts of a 3.5GHz Kaveri 7800.

The abnormaly lower score is possibly due to your single channel set up, Bristol Ridge 2MB L2 cache is quite small, the 200GE is less sensitive to channel count due probably to its larger cache.
 
  • Like
Reactions: amd6502

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Continuation of the FDSOI fiasco;

22FDX(+) APU -> 1H2020 aimed at Elkhart Lake bottom-end // Dual-core/Quad-core Tremont with GT1(32EU?)
12FDX CPU -> 2H2020(Early 2H) aimed at Jacobsville // 8-to-16 core TremontX.
12FDX APU -> 2H2020(Late 2H) aimed at Elkhart Lake top-end // Quad-core/Octo-core TremontX with GT2(64EU?)

Dali for AMD is aimed at the Y/U-series, Renoir is aimed at the U/H-series.

Investigating the long channel FDSOI 104PB10 for standard length and 116PB10 for long length. Should be a >1.35x increase in performance with a ~33% decrease in power.

A4-9120C 2.4 GHz/6W => ____C ~3.2 GHz/~4W
A6-9220C 2.7 GHz/6W => ____C ~3.6 GHz/~4W
There is more to it as well, as the FDSOI transistors have better leakage characteristics, increased lifespan expectancy, lower variability, etc.
 
Last edited: