AMD Bristol/Stoney Ridge Thread

[Justinbaileyman]

I feel like that these comments are based on the fact that the OEM versions are locked. But OEM =/= Self-builder Hardware. Unless the CPUs are already being pushed to their absolute maximum I dont see any reason to not have it be unlocked.
Zen is gonna beat up BR no matter what...so its not like they'd eat up their own shares by releasing K....the opposite is the case imho. Some cheap CPUs that let you play around until you get the real thing? Seems more attractive.

Thats what I'm saying!! (y)

 

[Justinbaileyman]

that was supposed to be a thumbs up sign

 

[The Stilt]

Voltage being higher than the FX mean nothing, comparatively it s like talking of a given volume of matter without specifying the density..

I find this statement quite amusing, especially when you have declared 14nm LPP superiority over Intel's processes, based on the high idle state voltage of a single Intel 14nm mobile product.

But no, really. The voltage is high even when compared to Bristol Ridge in general or even the same SKU itself. The ramp up in voltage beyond 3.5GHz is extremely rapid.

 

[Azuma Hazuki]

@The Stilt:

Why are they volted so high? I know, Ohm's Law, current = voltage/resistance, but still...it seems like AMD silicon comes with a massive overvolt ex-works. I rather hope AM4 motherboards can run a sort of quick stability test suite and undervolt and/or overclock the CPU when installed.

 

[The Stilt]

@The Stilt:

Why are they volted so high? I know, Ohm's Law, current = voltage/resistance, but still...it seems like AMD silicon comes with a massive overvolt ex-works. I rather hope AM4 motherboards can run a sort of quick stability test suite and undervolt and/or overclock the CPU when installed.

Probably due HDL, lower L2 latency and the process.
AMD parts are not factory overvolted any more than Intel products are. The load-line specification for both AM3+ and FM2+ is extremely loose (1.3 mOhm & 2.1mOhm) and because of that the voltage droop by the specification is very large. To ensure that the operation parameters remain within a spec, an AM3+ part which draws 100A of current and requires 1.3000V to be fully stable must have at least 1.4300V default voltage (130mV droop @ 100A). Some of the motherboards are built to have lower Rll (< 1.3mOhm / 2.1mOhm) than the specification dictates, or the end-user might adjust it to be lower than the default value. If the droop is lower or non-existing for either reason, it appears that the parts are extremely overvolted from the factory. The truth of course being that they are configured for perfectly right default voltage, which complies with the specifications (droop) and contains some standard safety margins.

On AM4 the situation is significantly better, since the load-line spec. is less than half of what it was on AM3+.

 

[Abwx]

I find this statement quite amusing, especially when you have declared 14nm LPP superiority over Intel's processes, based on the high idle state voltage of a single Intel 14nm mobile product.

But no, really. The voltage is high even when compared to Bristol Ridge in general or even the same SKU itself. The ramp up in voltage beyond 3.5GHz is extremely rapid.

If voltage was what define exclusively power dissipation then Bristol Ridge would had higher TDP than a FX4300, but of course you didnt think about this, instead you are changing the goal post by bringing Intel completely irrelevantly, although there can be a usefull comparison since their 14nm use higher voltage than their 22nm, yet it has somewhat better efficency despite the higher voltage, isnt it...

As for GF 14nm efficency in respect of Intel s there s no CPU using that process that would allow a comparsion, but perhaps that you have numbers that suggest that i would be wrong..?.

A hint, Zen consumed less than BDW in the Blender test, so at equal throughput it looks like GF s 14nm is certainly not lacking efficency, and that s a fact, not a theory like your non substancied statement that is even contradicted by the few info available.

P = C.V^2....

I see two terms here that have a say in power comsumption of a capacitive switching device, if voltage is higher by a ratio x then it can be compensated by a capacitance that is smaller by a ratio 1/x^2.

@The Stilt:

Why are they volted so high? I know, Ohm's Law, current = voltage/resistance, but still...it seems like AMD silicon comes with a massive overvolt ex-works. I rather hope AM4 motherboards can run a sort of quick stability test suite and undervolt and/or overclock the CPU when installed.

I explained it above, voltage in isolation has no meaning, that s just a red hering used by some people to badmouth AMD.

As for current in function of voltage that s still the formulae current = Voltage/resistance but resistance is defined as being equal to 1/2pi.F.C with C being the capacitance of the circuit and F the frequency, so it s current = Voltage.2.pi.F.C

 

[DrMrLordX]

We have seen some 845s at 4.5-4.6 GHz on air already (Flank3r tested one), and those are Carrizo, so I see no reason why AMD would have to lock down the A12-9800 on that account. They didn't lock the 7890k and it has practically 0 headroom as it is.

So I'm not 100% sure what's up with the A12-9800 unless AMD has plans to sell an unlocked product shortly thereafter.

That being said, we have no idea what AM4 is going to behave like after a bclk OC. FM2+ was kind of hit-or-miss.

I will only add that I was shocked by how much voltage AMD set as the default for GV-A1 chips on FM2+ until I started looking at power consumption numbers for myself. It definitely looked like the process tweaks had resulted in a lot of low-leakage parts compared to older KA-V1 chips. So despite the jacked-up voltage, most GV-A1 chips consumed less power overall, and of course they could be undervolted quite well depending on clockspeed.

 

[The Stilt]

We have seen some 845s at 4.5-4.6 GHz on air already (Flank3r tested one)

Got a link, I haven't seen one?

I've only seen some LN2 (high temp) / waterchiller + TEC runs in SuperPI at those clocks.

 

[Abwx]

We have seen some 845s at 4.5-4.6 GHz on air already (Flank3r tested one), and those are Carrizo, so I see no reason why AMD would have to lock down the A12-9800 on that account. .

Dunno for tests done by Flank3r but JagatOC overclocked their chip using the BCLK, it topped at 3.98GHz and comsumption didnt rise much, about 10W more at the APU level on top of the 46W/3.5GHz.

http://oc.jagatreview.com/2016/03/h...king-athlon-x4-845-carrizo-murah-tanpa-igp/5/

 

[Dresdenboy]

A hint, Zen consumed less than BDW in the Blender test, so at equal throughput it looks like GF s 14nm is certainly not lacking efficency, and that s a fact, not a theory like your non substancied statement that is even contradicted by the few info available.

This could be design related, as both AMD's M.P. and M.C. () stressed this point about power efficiency - even "floating point efficiency" in the SemiEngineering interview.

BDW itself is kind of the power efficient variant of HSW, but has 256b data paths, too. Look at my latest blog showing the diagram of HSW vs. IVB for the overall effect.

 

[DrMrLordX]

Got a link, I haven't seen one?

The only thing I can dig up right now is 4.2 GHz:

http://forum.hwbot.org/showthread.php?t=154372

I'm pretty sure I saw higher somewhere else but I'd have to do more digging. Oh wait . . .

http://hwbot.org/submission/3247009_namegt_cpu_frequency_athlon_x4_845_4585.33_mhz

Those volts are a bit creepy but it's there. I'd like to see the power consumption numbers on that OC.

 

[The Stilt]

The only thing I can dig up right now is 4.2 GHz:

http://forum.hwbot.org/showthread.php?t=154372

I'm pretty sure I saw higher somewhere else but I'd have to do more digging. Oh wait . . .

http://hwbot.org/submission/3247009_namegt_cpu_frequency_athlon_x4_845_4585.33_mhz

Those volts are a bit creepy but it's there. I'd like to see the power consumption numbers on that OC.

Yeah, but if they are just validations it is completely irrelevant. Validation requires "stability" of less than a second and is obviously completely irrelevant.
I have validated 32nm FX CPUs at 6GHz on air and at 8.7GHz on LN2 and obviously these results have no correlation what so ever with the actual capabilities of the CPUs in normal conditions.
I've probably grown old and nowdays I'm not personally interested in any other kind of stability but Prime95 proof

 

[PhonakV30]

6GHz on air !! damn.please give me a hot cake!

 

[DrMrLordX]

Yeah, but if they are just validations it is completely irrelevant. Validation requires "stability" of less than a second and is obviously completely irrelevant.
I have validated 32nm FX CPUs at 6GHz on air and at 8.7GHz on LN2 and obviously these results have no correlation what so ever with the actual capabilities of the CPUs in normal conditions.
I've probably grown old and nowdays I'm not personally interested in any other kind of stability but Prime95 proof

Well I'm pretty sure Flank3r ran SuperPi on his 845 @ 4.2 GHz, not sure with the other guys though. It would be fun to play with one if I had spare cash and an extra system . . . but truth be told, we'll learn more in October from those AM4 systems, and the main event comes in February. So it's academic at this point.

 

[The Stilt]

6GHz on air !! damn.please give me a hot cake!

Actually it was on AIO (Asetek 570LC) now that I've checked it.
But anyway, irrelevant for determining the CPU potential all the same.

 

[KTE]

Well, I turned away from desktops circa 2012. My work and it's two main partners has done the same since 2015.

But now, I'm back in for a desktop for good and I'm looking to test BR 9800 vs. Llano 3870 on the CPU side.

Altho Zen or KL/CL is what I'll settle with.

Sent from HTC 10
(Opinions are own)

 

[dooon]

HP Pavilion Desktop - 510-p101np PC Product Specifications
http://support.hp.com/ca-en/document/c05254833/
Motherboard : Willow
Processor : AMD A10-9700 (Bristol Ridge)
HP Desktop PCs - motherboard specifications, Willow
http://support.hp.com/ca-en/document/c05254568
Chipset : AMD Promontory FCH
Socket type : AM4

 

[cbn]

HP Pavilion Desktop - 510-p101np PC Product Specifications
http://support.hp.com/ca-en/document/c05254833/
Motherboard : Willow
Processor : AMD A10-9700 (Bristol Ridge)
HP Desktop PCs - motherboard specifications, Willow
http://support.hp.com/ca-en/document/c05254568
Chipset : AMD Promontory FCH
Socket type : AM4

Interesting this uATX board "Willow" has the Promontory FCH (located on the lower right side of board) when it only has two SATA ports and a single PCIe x 16 slot and M.2 (for wireless). Does anyone know why that is?

(I would have expected this to be a chipset-less (FCH-less) board)

P.S. Also note this board only lists support for 65W processors.

 

[DrMrLordX]

That is a bit weird, though the 65W max is probably normal. That's the TDP for the top-end Bristol Ridge anyway (A12-9800).

 

[cbn]

Looking more carefully at the HP AM4 "Willow" motherboard specs in post #392 I notice it does have usb 2.0 which is not included in the Bristol Ridge SoC I/O shown below:

However, if that usb 2.0 wasn't included what else would have been sacrificed if HP went SoC only for that board (shown below):

http://support.hp.com/ca-en/document/c05254568

Would the PCIe x 16 slot be running at x8 or slower if the above board was a "SoC only" one (without usb 2.0)?

 

[majord]

Benchmarks!

seemingly DDR4 is @ 2133

http://www.bodnara.co.kr/bbs/article.html?num=134612&mn=4

 

[DrMrLordX]

Not bad. Unfortunately, it looks like BR has the expected half-sized l2 cache of Carrizo. That's going to be a problem with some games/applications.

 

[Justinbaileyman]

2 x 1MB L2?? this is getting worse and worse

 

[Azuma Hazuki]

The only reason I can see them doing that is so as not to cannibalize Zen sales. BR is coming at a weird time; it's basically a stopgap for the low-mid end of the market, as Zen-based Ravenridge APUs won't be ready for a while, but at the same time, it may very well chew on the lower side of the Zen CPU's marketshare. I don't envy AMD right now...

 

[deasd]

Nice to see first AM4 board that is not from OEM. No surprise for CPU side but IGP and overall efficiency are quite good. If it price a bit under 95w 7890K it would be a good buying.
Wait to see more about different DRAM config comparison in game, since DDR4 could vary from 2133 to 3600Mhz......