AMD Bristol/Stoney Ridge Thread

[USER8000]

512SP @ 757MHz and 1866MHz max. on FX-9800P
I didn't think 1866MHz DDR4 was even available

Emm,why are you showing such surprise:

http://wccftech.com/amd-bristol-ridge-fp4-family-leak/
http://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series for Notebooks FX-9800P.html

That was leaked months ago - the FX9800P is a 15W TDP part.

The base clockspeed for the CPU is 600MHZ higher than the previous chip which is huge.

Also,you might have noticed something about the FX8800P - it is configurable between 12W to 35W TDP,so its most likely those clockspeeds quoted are at a higher TDP.

http://www.notebookcheck.net/AMD-FX...r-Specifications-and-Benchmarks.144074.0.html

The AMD FX-8800P is the top-tier Carrizo APU of 2015 for mainstream laptops. It is a SoC containing two Excavator CPU modules (with 4 integer and 2 FP units counted as 4 compute cores), a Radeon R7 graphics card with 8 compute cores (512 shaders), a DDR3-2133 (35 Watt cTDP) memory controller and a video engine (including UVD 6 with hardware HEVC / H.265 decoding).

The performances of the CPU and GPU cores depend on the configurable TDP. This cTDP can range from 12 to 35 Watts depending on the cooling solution and size of the laptop. Thus, clock speeds of the compute cores will vary. E.g., the GPU can score between 1992 points (15 Watts) to 2753 points (+38% with 35 Watts cTDP) in 3DMark 11 (P).

According to AMD, the 35 Watt configuration scores 12% (Cinebench single-thread) and 15% (Cinebench Multi-threaded) higher compared to the previous Steamroller generation (most likely the AMD FX-7600P). If configured to 15 Watt instead, the gains are higher (40% to 55%) compared to the 19 Watt Kaveri FX-7500 APU.

2133MHZ DDR3 was only supported by the FX8800P when it was configured to a 35W TDP.

The FX9800P clockspeeds are only quoted at a 15W TDP.

 

[AtenRa]

Yea the 9700B at 15W TDP should be the better choice for perf/$ vs the 9800B.

And for the millionth time, please please please release a f.....king 35W TDP with out an idGPU.

 

[USER8000]

Yea the 9700B at 15W TDP should be the better choice for perf/$ vs the 9800B.

And for the millionth time, please please please release a f.....king 35W TDP with out an iGPU.

Remember a while back,the Anandtech article said OEMs were configuring the 35W parts at under 20W,so in reality the FX8800P is probably running at lower than stated clockspeeds.

It does seem,that AMD decided to make the FX9800P a 15W part from the beginning instead of having it a variable TDP,so if those clockspeeds hold true at 15W,then it should be quite a big improvement over the FX8800P at under 20W.

The main concern was with OEMs only running the CPU in single channel,and looking at that previous post,the FX9800P was running in dual channel,which means the motherboards can support it and OEMs are using dual memory modules.

That is the bigger deal here.

 

[Abwx]

The main concern was with OEMs only running the CPU in single channel,and looking at that previous post,the FX9800P was running in dual channel,which means the motherboards can support it and OEMs are using dual memory modules.

That is the bigger deal here.

Not only that s not a deal at all but this is an urban legend born from the fact that AT tested non commercial samples that had single channel for some but the actual commercial prodution is dual channel, only exemple that had single channel was an unavailable laptop from Dell used as exemple ad nauseam by the usual crowd...

So far that s the lowest end Carrizo using a 2C at Lenovo and it still has a dual channel :

http://www.cnet.com/products/lenovo...6-8500p-4-gb-ram-500-gb-hdd-20ey000aus/specs/

Now for whom want to say the contrary here the Carrizo laptops sold in Germany, can someone point me if there s a single channel laptop in the list..?. :

https://geizhals.de/?cat=nb&xf=9690_Carrizo#xf_top

https://geizhals.de/?cat=nb&xf=9690_Carrizo+Pro#xf_top

The Bristol Ridge based :

https://geizhals.de/?cat=nb&xf=9690_Bristol+Ridge#xf_top[/QUOTE]

 

[The Stilt]

Remember a while back,the Anandtech article said OEMs were configuring the 35W parts at under 20W,so in reality the FX8800P is probably running at lower than stated clockspeeds.

It does seem,that AMD decided to make the FX9800P a 15W part from the beginning instead of having it a variable TDP,so if those clockspeeds hold true at 15W,then it should be quite a big improvement over the FX8800P at under 20W.

The main concern was with OEMs only running the CPU in single channel,and looking at that previous post,the FX9800P was running in dual channel,which means the motherboards can support it and OEMs are using dual memory modules.

That is the bigger deal here.

There never was > 15W Carrizo SKUs outside R & G series.
FX-8800P has TDP of 15W (25W during boost), but it supports cTDP up to 35W (42W during boost).
FX-9800P has the same 15W / 25W TDP, but the cTDP can only be configured downwards (12-15W).

 

[raghu78]

sorry wrong thread.

 

[USER8000]

I didn't realise that Abwx,but AMD did do that article with Anandtech,so its a bit of a failure on their side for sort not pushing out representative hardware.

There never was > 15W Carrizo SKUs outside R & G series.
FX-8800P has TDP of 15W (25W during boost), but it supports cTDP up to 35W (42W during boost).
FX-9800P has the same 15W / 25W TDP, but the cTDP can only be configured downwards (12-15W).

That is the point though as noted by Notebookcheck,the clockspeeds for the FX8800P are at a 35W TDP,not configured to 15W,as they mention GPU clockspeeds are much lower than stated at 15W,and say the score is 38% higher going from 15W to 35W.

If you look at the FX9800P it is rated at 15W TDP only with higher options not available for the chip.

This is why,the boost in base clockspeed of the CPU by 600MHZ looks quite big,as is the GPU clockspeeds looking almost the same as the FX8800P in 35W mode.

This does not preclude the FX9800P massively throttling either,but if they can maintain those clockspeeds,it might be a good uplift over the FX8800P in 15W mode.

 

[DrMrLordX]

Numbers are accurate,

You're completely missing the point. The only way someone is going to report a motherboard failure from running a 4m Vishera in a 4+1 phase motherboard is if they manage to pop a VRM and the board goes down. If the thing throttles they'll report faulty CPU. Mb failure rates are essentially meaningless.

There are loads of crap AM3+ boards that have 4m chips on their compatibility list that will do a poor job (at best) of running those chips. Particularly 760G boards. Ugh. They aren't going to "fail", they'll induce throttling.

You can keep posting hardware failure rates, but it has nothing to do with what The Stilt was saying, nor does it contribute to the conversation in any meaningful way.

 

[Abwx]

You're completely missing the point. The only way someone is going to report a motherboard failure from running a 4m Vishera in a 4+1 phase motherboard is if they manage to pop a VRM and the board goes down. If the thing throttles they'll report faulty CPU. Mb failure rates are essentially meaningless.

There are loads of crap AM3+ boards that have 4m chips on their compatibility list that will do a poor job (at best) of running those chips. Particularly 760G boards. Ugh. They aren't going to "fail", they'll induce throttling.

You can keep posting hardware failure rates, but it has nothing to do with what The Stilt was saying, nor does it contribute to the conversation in any meaningful way.

What doesnt contribute to the discussion is that some people make conclusions while ignoring the numbers.

For one a 95W TDP FX will not throttle on a 95W MB, for the very simple reason that no 95W TDP FX use 95W even under Prime 95.

FTR under Prime 95 the comsumption for the FX8370E is 80W and about 65W with regular MT loads like CB or Fritzbench, so there s no way that chips could throttle in a 95W MB other than in the imagination of some people.

Moreover even some 125W chips will work flawlessly without throttling or even exceding 95W, exemple is the FX4350 wich use 80W under Prime 95 and 61W in usual MT loads.

Now that s numbers that are sourced, contrary to general statement that come more or less from nowhere, or from urban legends at best.

 

[DrMrLordX]

What doesnt contribute to the discussion is that some people make conclusions while ignoring the numbers.

Keep posting irrelevant numbers and I'll keep ignoring them. Besides, AMD only recently released a 95W 4m Vishera. Over the vast majority of the product's lifespan, the minimum TDP of a 4m Vishera chip was the 8320 which is listed as 125W TDP. 4+1 power phase boards are generally inappropriate for such processors at default settings where the majority of users would run them.

And if you want to bring up boards with "good" VRMs in 4+1 configurations, well, it's 2016 and we're still getting this:

http://www.overclock.net/t/1610853/voltage-throttling-are-my-mosfets-overheating

Yes, that's right, a guy with an old 8120 (TDP 125W) running on a cheap-ish Asus board that is suffering from VRM throttling. If you look at the supported CPU list, even the 8370 is supported. Really? Okay, whatever.

This problem is no "urban legend"; here's an example of VRM throttling staring you in the face.

AMD can't afford any more bs motherboards that fail to meet power requirements. In this case it's the motherboard manufacturers' fault for drafting such ridiculous CPU compatibility lists. Hopefully they require AM4 boards to be built to exacting specifications for proper power delivery for every supported product. If they need to segment based on chipset then fine, at least the OEMs will have the advantage of being able to implement a common socket across multiple chipsets even if not all chipsets will support all AM4 CPUs according to how OEMs implement them.

 

[Exist50]

On the topic of a proper AM1 replacement, I do wonder if it would not be possible for an OEM to take one of their lower end mITX or μATX and cut it down to the bare minimum, including cutting the power delivery, to make a board only compatible with the 35W Bristol Ridge chips for very cheap. There's a pretty decent span of an upgrade path within even the 35W range, and it'd certainly be suitable in performance for a family PC or HTPC. In particular, sticking to the bare essentials provided by the integrated X/B/A chipset would be more than fine. Two SATA for mass storage and/or disk drive, and an M.2 slot for a boot drive. The x8 PCIe would be plenty anyways, as would the 4 USB ports. Now, of course it'd still be a bit more expensive than AM1 in every regard, but there should be some scaling advantages from sharing some of the design from another product. As far as pipe-dream stuff, USB C for power would be great, albeit a bit expensive to implement.

Of course, such a product would be a nightmare for marketing, so probably won't exist unless Asrock does something crazy again like they used to. I wonder why AMD even bothered advertising AM4 as an AM1 replacement. They could easily have left it off the slides and no one would have thought twice about it.

 

[AtenRa]

I wonder why AMD even bothered advertising AM4 as an AM1 replacement. They could easily have left it off the slides and no one would have thought twice about it.

Where did you see AMD advertising AM4 as AM1 replacement ??

 

[Exist50]

Where did you see AMD advertising AM4 as AM1 replacement ??

This.

 

[cbn]

X/B/A300? "Enabling the smallest form factors with the ultimate power and space efficiency"

I wonder what that handles? Is this "chipset" really SoC only with no actual chipset? Limited TDP?

 

[cbn]

For "TBA" listed across from Enthusiast (and corresponding with 990FX and A88X), I wonder if that would eventually be for a higher than 95W TDP specification?

 

[cbn]

On the topic of a proper AM1 replacement, I do wonder if it would not be possible for an OEM to take one of their lower end mITX or μATX and cut it down to the bare minimum, including cutting the power delivery, to make a board only compatible with the 35W Bristol Ridge chips for very cheap.

Why not just use BGA then? This way AMD doesn't have split their processor dies between two platforms (BGA mobile and AM4).

There's a pretty decent span of an upgrade path within even the 35W range, and it'd certainly be suitable in performance for a family PC or HTPC. In particular, sticking to the bare essentials provided by the integrated X/B/A chipset would be more than fine. Two SATA for mass storage and/or disk drive, and an M.2 slot for a boot drive. The x8 PCIe would be plenty anyways, as would the 4 USB ports. Now, of course it'd still be a bit more expensive than AM1 in every regard, but there should be some scaling advantages from sharing some of the design from another product. As far as pipe-dream stuff, USB C for power would be great, albeit a bit expensive to implement.

Of course, such a product would be a nightmare for marketing, so probably won't exist unless Asrock does something crazy again like they used to. I wonder why AMD even bothered advertising AM4 as an AM1 replacement. They could easily have left it off the slides and no one would have thought twice about it.

Actually the performance difference between the three 35W quad core Bristol Ridge mobile APUs (FX-9830P, A12-9730P, A10-9730P) is fairly small (much smaller than the Intel 35W range which which spans Celeron G3900T (dual core w/o HT, 2MB cache, GT1) to Core i7 6700T (quad core w/ HT, turbo, 8MB cache, GT2)).

With that mentioned perhaps they intend to use Stoney Ridge for a 25W to 35W platform?

 

[AtenRa]

This.

Ok, thought you were saying a direct replacement. Well AM4 will span from entry-level dual cores to 8-Core 16Threads CPUs, in that regard it will also replace the AM1. But we have to wait and see how they will do that.

 

[dooon]

Bristol Ridge "Athlon X4 950" (AM4) photo
https://gigglehd.com/gg/hard/311514

 

[cbn]

Ok, thought you were saying a direct replacement. Well AM4 will span from entry-level dual cores to 8-Core 16Threads CPUs, in that regard it will also replace the AM1. But we have to wait and see how they will do that.

I am assuming processors meant to work in X/B/A300 will work in all the other AM4 chipset boards.....but not all AM4 processors will work in X/B/A300:

If true, that should be interesting. But where will the TDP limit be? 65W? or even lower than that?

65W would be great because we could have a budget board for 65W 4C/8T Summit Ridge and 65W A6-9500* Bristol Ridge (as well as any future harvested dual core Zen APU).

*The first AMD dual core APU (based on the big iGPU die) with more than half its graphics core enabled.

 

[cbn]

Another thing to wonder about is whether X/B/A300 will be split up into the following designations:

X300
B300
A300

.....With each designation signifying a different form factor range, TDP limit and set of motherboard features?

 

[Exist50]

Another thing to wonder about is whether X/B/A300 will be split up into the following designations:

X300
B300
A300

.....With each designation signifying a different form factor range, TDP limit and set of motherboard features?

Since it seems to point towards the integrated chipset's features, I wonder if that's for the integrated of Bristol Ridge, Summit Ridge, and Raven Ridge. Maybe A=BR, B=RR, and X=SR?

 

[cbn]

Since it seems to point towards the integrated chipset's features, I wonder if that's for the integrated of Bristol Ridge, Summit Ridge, and Raven Ridge. Maybe A=BR, B=RR, and X=SR?

Yes, I agree X/B/A300 boards are likely SoC only (ie, they have no external chipset (aka I/O expander) like A320 or B350 (and the TBA one)).

And could it be the X300 has the highest TDP with the largest form factor and the highest amount of SoC I/O available on the board (eg, PCIe x 16 slot and SATA ports), while A300 might be targeted for a NUC type form factor with lower TDP and I/O (eg, lack of PCIe x 16 slot and perhaps SATA ports, but M.2 for SSD)?

 

[cbn]

In post #319, I mentioned a 65W limit for X/B/A300......but maybe the limit could be 95W? This way the X300? Could accommodate the highest TDP Summit Ridge, the B300 could accommodate the 65W processors, and A300 could accommodate only the lowest TDP processors (Stoney Ridge?)

 

[cbn]

My apologies for missing the slide below (posted earlier in the thread) which implies X/B/A300 as SoC only :

 

[cbn]

So assuming X300 is the top "chipset-less" motherboard (and is capable of Supporting the 95W Summit Ridge), does it come with power planes for iGPU or not?

I'm actually thinking it may not. (re: APUs are rated for 65W and below, so B300 might be the highest power "chipset-less" motherboard to have power planes to iGPU).

P.S. Another thing to consider (which might support the idea of "no power planes to iGPU" for 95W X300) is that these chipset-less motherboards are intended for SFF.......and perhaps having power planes/VRMs to iGPU on a 95W X300 competes with space needed for power delivery to CPU?