AMD Bristol/Stoney Ridge Thread

[Abwx]

So basically you claim that AMD has improved the performance per watt of Bristol Ridge (particularly with A12-9800) by > 60%?
62.6% improvement would be required to sustain the advertized clocks, even if the voltage would be identical as on Carrizo at 3400MHz CPU and 800MHz GPU frequencies.
Since they won't be (A12-9800 runs up to 1.5V VDD_C at default), I'm being extremely generous with these estimations

45W for 3.4GHz (+11.8%) and 40W for 800MHz (+38.5%) GPU figures used.

Also on Carrizo and newer, the CPU and GPU are not the only domains which take their cut from the total TDP budget

I didnt claim what you said and btw they dont need 60%, from the curves they published they gained something like25- 30% in favourable parts of the curve, rest is improvement of their AVFS and power management, at 30% lower actual TDP that s enough to get in the 65W official TDP rating..

And about voltages they are comparable only for a same process, otherwise you have to take account of the difference in switching capacitance to have an idea of the respective efficencies, for instance Intel s 14nm use higher voltage/frequency than their 22nm but it is overcompensated by the reduction of switching capacitance, hence the (slightly) better perf/watt of the former.

 

[cbn]

Do we know if the AM4 boards will be able to operate without a chipset? I am specifically thinking of Summit Ridge.

 

[DrMrLordX]

If by chipset you mean FCH, then . . . it may depend on the chipset.

 

[The Stilt]

Neither Bristol, Summit Ridge or Raven should require the external chipset. Promontory is basically a IO expander, adding functionality which isn't included in the internal FCH / SCH.

 

[VirtualLarry]

Neither Bristol, Summit Ridge or Raven should require the external chipset. Promontory is basically a IO expander, adding functionality which isn't included in the internal FCH / SCH.

So we could see AM4 mobos with limited expansion capabilities for as cheap as AM1 motherboards? ($30 or less.)

 

[The Stilt]

So we could see AM4 mobos with limited expansion capabilities for as cheap as AM1 motherboards? ($30 or less.)

Not a chance.

 

[Shehriazad]

I hope that the 960K/970K isn't gonna

Not a chance.

Eh...FM2+ Boards exist for $40 (or less, depending on region)....so I don't see why there's "not a chance". Care to elaborate?

 

[The Stilt]

I hope that the 960K/970K isn't gonna


Eh...FM2+ Boards exist for $40 (or less, depending on region)....so I don't see why there's "not a chance". Care to elaborate?

AM4 requires much beefier VRMs (all CPU / APU planes), higher quality PCB (due higher currents, faster signaling), etc. The design guidelines for all Zen based platforms are quite demanding, so you cannot get away with the same garbage quality as one could with "AM1" or FM2+. Unless you want to break the platform cross compatibility of course. All FM2+ boards < 60$ (CSP) are generally garbage and in most cases cannot operate properly with all the APUs available for the platform.

 

[AtenRa]

There gonna be 35W TDP BristolRidge APUs and they need beefier VRMs than 65W TDP FM2+ ???

 

[The Stilt]

There gonna be 35W TDP BristolRidge APUs and they need beefier VRMs than 65W TDP FM2+ ???

Obviously not, but would you purchase a AM4 motherboard which is only compliant with certain AM4 CPU/APUs?
That basically happened with AM3+ and somewhat on FM2+ and it gave the CPU & APUs a worse reputation than they deserved.

 

[AtenRa]

Obviously not, but would you purchase a AM4 motherboard which is only compliant with certain AM4 CPU/APUs?
That basically happened with AM3+ and somewhat on FM2+ and it gave the CPU & APUs a worse reputation than they deserved.

If it is cheap and clearly states its only for UP TO 65W TDP APUs yes.

Those AM3+ motherboards you are talking about where labeled as 125W TDP when in reality they where simple only up to 95W TDP.

An entry level chip-less up to 65W TDP motherboard will be fine for the vast majority of users.

 

[DrMrLordX]

Rest assured that if cheaper common-socket products are available, some jackass is going to try to peddle them to people that want something more. It's the same deal as what happened with Carrizo . . . give them Carrizo-L and Carrizo board specs and make them pin-compatible, and you wind up with Carrizo soldered to Carrizo-L spec boards.

AMD is going to have to take a stand on quality if they don't want more 4+1 phase VRM-type fiascos like they had on AM3+, not to speak of some of the cheap crap FM2+ boards out there. Hell even some of the expensive FM2+ boards sucked. There are still people struggling with the Gigabyte Sniper boards! Those things are terrible. Why did they have to be so bad when the UP4 was so good?

 

[jpiniero]

Knowing what OEMs love to do with AMD products, if it's possible to not have a chipset they will not use it.

 

[Abwx]

The design guidelines for all Zen based platforms are quite demanding, so you cannot get away with the same garbage quality as one could with "AM1" or FM2+. Unless you want to break the platform cross compatibility of course. All FM2+ boards < 60$ (CSP) are generally garbage and in most cases cannot operate properly with all the APUs available for the platform.

So a good day for you is a day where you re badmouthing AMD using urban legends..?..

Let s check the reliability and see what is actually the real garbage rather than throwing the usual fud, isnt it, that s the defect rate of the retailer that own Hardwre.fr :

http://www.hardware.fr/articles/947-2/cartes-meres.html

• • 4,55% X99
  • 3,36% Z170
  • 2,09% Z87/Z97
  • 2,20% H87/H97
  • 1,48% B85
  • 1,03% H81

• 3,89% 990
• 3,10% 970
• 1,68% A88X
• 1,36% A78
• 2,33% A58/A55

I bolded the chipsets that are currently used in about all mainstream and low cost cards, we can see that the expensive X99 MBs are the less reliable, and in the mainstream AMD MBs are more reliable than Intel s..

 

[superstition]

Those rates have some unfortunate variables, though — like reliance upon the end user to identify the problem and return the board. Different products mean different user bases and thus different group behaviors. More demanding vs. less. More knowledgeable versus less.

Just the fact that all FX chips have unlocked multipliers could make a difference in perceived issues, versus trying to use BCLK to overclock. However, Intel's comparatively rapid socket changes could lead to lesser QC. Broadwell C was given rather short shrift by board makers it seems, for instance, particularly in terms of overclocking.

 

[superstition]

If it is cheap and clearly states its only for UP TO 65W TDP APUs yes.

Those AM3+ motherboards you are talking about where labeled as 125W TDP when in reality they where simple only up to 95W TDP.

An entry level chip-less up to 65W TDP motherboard will be fine for the vast majority of users.

The "Fatality" AsRock 970 board, which doesn't even have LLC, was supposed to support 9000 series CPUs. Good luck with that, even with a fan on the VRM sink.

It's correct that AMD needed to take action to tighten QC for motherboards. It has been too much of a Wild West situation. Improving transparency for important specs for the consumer is similarly important. Nonsense like "military grade" should be replaced with a clear understanding of what the board's power delivery is capable of.

All the lies about phases should be dealt with, especially. I looked at Micro Center at reviews for the UD3P 970 board and three of like eight comments praised the "8 phase" VRMs. Boards that use doublers need to be labeled as such.

 

[superstition]

AM4 requires much beefier VRMs (all CPU / APU planes), higher quality PCB (due higher currents, faster signaling), etc. The design guidelines for all Zen based platforms are quite demanding

This is great news. What about the issue of using low-grade Nikos MOSFETs?

 

[cbn]

For a AM4 board without chipset, do we know what I/O the Summit Ridge processor comes with?

 

[The Stilt]

For a AM4 board without chipset, do we know what I/O the Summit Ridge processor comes with?

Comes with?
You mean the internal FCH / SCH (i.e Taishan)?

 

[PhonakV30]

Dual channel DDR4 is enabled on the Aspire E5-553G
https://www.reddit.com/r/Amd/comments/51g65p/dual_channel_ddr4_is_enabled_on_the_aspire_e5553g/

http://valid.x86.fr/7y7q6f

 

[The Stilt]

Dual channel DDR4 is enabled on the Aspire E5-553G
https://www.reddit.com/r/Amd/comments/51g65p/dual_channel_ddr4_is_enabled_on_the_aspire_e5553g/

http://valid.x86.fr/7y7q6f

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

 

[PhonakV30]

according to CPU-z ,

Total Size :16384 MB
Type : Dual Channel (128 bit) DDR4-SDRAM
Frequency : 931.4 MHz - Ratio 3:28
Timings : 13-13-13-31-44 (tCAS-tRC-tRP-tRAS-tCS)
Slot #1 Module :
Hyundai Electronics 8192 MB (DDR4-2137) - P/N: HMA41GS6AFR8N-TF
Slot #2 Module :
Hyundai Electronics 8192 MB (DDR4-2137) - P/N: HMA41GS6AFR8N-TF

So it's SK Hynix
https://www.skhynix.com/products.view.do?vseq=1542&cseq=75

and According to Sk Hynix sheet : Part Number : TF ==> 2133 15-15-15
So IMC can't handle it well for 2133Mhz instead of 1866Mhz?

 

[The Stilt]

No, 1866MHz is the maximum allowed speed on FX-9800P.

 

[DrMrLordX]

So a good day for you is a day where you re badmouthing AMD using urban legends..?..

Let s check the reliability and see what is actually the real garbage rather than throwing the usual fud, isnt it, that s the defect rate of the retailer that own Hardwre.fr :

The motherboard defect rate may not cover circumstances where a purportedly-supported CPU fails to operate normally due to throttling from overheating VRMs. I would expect most users to report the CPU as being problematic in those situations.

You know what The Stilt is talking about here, and trying to obfuscate the situation by talking about mb failure rates is disingenuous at best.

 

[Abwx]

The motherboard defect rate may not cover circumstances where a purportedly-supported CPU fails to operate normally due to throttling from overheating VRMs. I would expect most users to report the CPU as being problematic in those situations.

You know what The Stilt is talking about here, and trying to obfuscate the situation by talking about mb failure rates is disingenuous at best.

Numbers are accurate, that s MBs whose defect has been established as originating from the MB by the retailer, they say that there can be a few wich are discutable but overall Hardware.fr state that the numbers are accurate, so much for being disingenious, the numbers are what they are and certainly that they contradict some urban legend spreaders...

Indeed they point that the MBs which are most overclocking oriented are the ones wich have higher return ratios, but even in that case the AMDs have low returns.

For infos here the return rates by brand and chipsets :


Worst ratio MBs :

• 7,76% ASUS Rampage V Extreme X99
• 7,50% MSI Z97 Gaming 7
• 5,63% ASROCK 980DE3/U3S3 R2.0
• 5,02% MSI 970 Gaming
• 4,90% ASUS Z170-A

By MB brand for Intel X99

• 7,76% ASUS Rampage V Extreme
• 3,95% Gigabyte GA-X99-UD4
• 0,71% ASUS X99-A

By MB brand for Intel Z170/Z97

• 7,50% MSI Z97 Gaming 7
• 4,90% ASUS Z170-A
• 4,79% ASUS Z170 PRO Gaming
• 4,32% MSI Z97-G45 Gaming
• 3,00% ASUS Maximus VII Formula Z97
• 2,19% MSI Z97 Gaming 3
• 2,05% ASRock Z97 Anniversary
• 2,02% MSI Z97 Gaming 5
• 1,89% ASUS Z97-A
• 1,85% ASUS Maximus VII Ranger Z97
• 1,84% MSI Z97 PC MATE
• 1,83% ASUS Maximus VII Hero Z97
• 1,75% ASUS Z97I-PLUS
• 1,72% ASUS Z97 PRO Gamer
• 1,28% Gigabyte Z97X-Gaming 3
• 1,25% ASUS Z87-PRO (V Edition)
• 1,19% ASUS Z97-P
• 1,14% Gigabyte Z97-D3H
• 1,13% ASUS Z97-K
• 1,02% ASUS Z170-P D3

Intel by chipset :

• 4,55% X99
• 3,36% Z170
• 2,09% Z87/Z97
• 2,20% H87/H97
• 1,48% B85
• 1,03% H81

AMD by chipset :

• 3,89% 990
• 3,10% 970
• 1,68% A88X
• 1,36% A78
• 2,33% A58/A55