piledriver - Official AMD FX4300,FX6300 and FX8350 review thread

Marine-RX179 · 29 Oct 2012 at 10:42

passey89 said:
will let you know.

I have full water loop ready for my 8350 as soon as OCUK get it in stock.

Hoping for 5Ghz.

Think the reason for the much higher power-consumption at beyond 4.6GHz is most likely due to much more aggressive vcore is needed for the top-end overclocking at beyond 4.6GHz.

But a good board would definitely help as well...may be one of the Gigabyte with their latest Ultra Durable 5/VRM enhanced tech? It allows same high overclock of IvyBridge i5 to 4.7GHz but at a low 1.30v, comparing to their own Ultra Durable 4 boards and other brand boards which require 1.37-1.38v on average, resulting in good 8C lower...

But then again, that would involves getting a new board...

Davedree · 29 Oct 2012 at 10:57

passey I may be wrong but that the asrock 990fx extreme 3 only has a a 4+1 vrm for the power without a heatsink, so they might run hot/hold back you overclock with instability when you try and push the chip past its thermal limits. Ideally If i was investing in an am3+ back when they were coming out I would made sure it was 8+2. But i made a better choice and jumped ship and I don't regret it at all

Martini1991 · 29 Oct 2012 at 10:58

humbug said:
Running Phenom II x6's at 4.2Ghz is not the norm, 45nm with 6 cores they can be tricky to get up to that speed for 24/7 use.
4Ghz is the usual for them

They are also missing some modern features and instruction sets that the Piledriver chips do have.
Phenom's are old, they are older than the first i7 chips.

I'm watching a couple of threads on other forums in which people are have taken possession on FX-8350 to replace Thubans and Bulldozers.

So far one stable at 5Ghz on water and another at 4.7Ghz on air, they clock quite a bit higher than Bulldozer and still run cooler. at that they are some way faster than an x6 @ 4 / 4.2Ghz.

Phenom II = 2009.
i7 920 = 2008.

Replacing a Bulldozer with a Piledriver, meh. (While it may be the most cost effective at that moment, it's only due to their choices previously, overall their performance has cost them more than an Intel set up in Q1 2011.)

4.2GHZ on Thuban's was pretty common on launch, the 1090T's etc, I'd say 4.2GHZ was a viable go to goal on Phenom II's given their maturity now.

And I can't see why anyone would switch from a Thuban to an FX8350, it wouldn't gain much unless they were encoding heavily constantly.

Although I don't see that clocking much higher than Bulldozer, as they were doing 4.8GHZ and could be stable at 5GHZ.

As for the instruction sets, it's not like the Phenom II's are prevented from running stuff that we'd care about due to instruction sets, Bulldozer had a much more comprehensive list of instruction sets, didn't help it much (Except for some occasions with AVX)

One thing I've noticed in 4 years.
People will throw more money at AMD than Intel, then claim Intel are more expensive, I did it myself till I realised it was a illogical.

wolvers · 29 Oct 2012 at 10:58

Reading some of the early reports from 8350 users overclocking, the gigabyte boards aren't ideal because of the way that they apply increased voltage.

Also, the concensus seems to be that they respond better to fsb increase than multipliers. Lower vcores needed etc.

ntg · 29 Oct 2012 at 10:59

Is there a decent performance improvement going from 8120 to 8320? I couldn't find any benchmarks anywhere for these specific cpu comparison.

Don't need the upgrade that much but I'd rather sell my 8120 and put a bit now to upgrade than having to shell out the whole retail price next year for another cpu when mine becomes dirt cheap in the second hand market.

Hagar · 29 Oct 2012 at 11:18

wolvers said:
Reading some of the early reports from 8350 users overclocking, the gigabyte boards aren't ideal because of the way that they apply increased voltage.

Also, the concensus seems to be that they respond better to fsb increase than multipliers. Lower vcores needed etc.

This was always the way with AMD unlocked or modded cpu's. Find the highest system clock from memory and mother board , set voltages, then use multipliers to get the highest cpu clock, possibly backing off the system clock a little to get the next highest multiplier increment. I always use a higher clock than 200MHz through habit and back off one multiplier increment for 24/7 usage.

Martini1991 · 29 Oct 2012 at 11:22

Yeah, boards like the Crosshair IV could get the most out of a chip, you could lessen the need for higher multiplier as you could run a 350MHZ bus.

humbug · 29 Oct 2012 at 11:27

nkata said:
This was always the way with AMD unlocked or modded cpu's. Find the highest system clock from memory and mother board , set voltages, then use multipliers to get the highest cpu clock, possibly backing off the system clock a little to get the next highest multiplier increment. I always use a higher clock than 200MHz through habit and back off one multiplier increment for 24/7 usage.

It is, as always with AMD a good overclock consist of a combination of HT-Bus (FSB) and the multiplier + NB multiplier + NB Volts and VDDA Volts on Piledriver.

And to think the guy from Overclockers.net in that youtube twice linked in this thread now; spent a good 5 minutes ranting about the fact that he could not adjust the HT-Bus (FSB) at all.
I have jet to see anyone have the same problem :rolleyes:

Martini1991 · 29 Oct 2012 at 11:30

humbug said:
It is, as always with AMD a good overclock consist of a combination of HT-Bus (FSB) and the multiplier + NB multiplier + NB Volts and VDDA Volts on Piledriver.

And to think the guy from Overclockers.net in that youtube twice linked in this thread now; spent a good 5 minutes ranting about the fact that he could not adjust the HT-Bus (FSB) at all.
I have jet to see anyone have the same problem

1.) Isn't Overclockers.net

2.) No he didn't say that.

He did have a moan about what he could take the bus speed too (HT Reference clock) however, although I'm sure that's more down to the board than the chip.

opethdisciple · 29 Oct 2012 at 11:34

ntg said:
Is there a decent performance improvement going from 8120 to 8320? I couldn't find any benchmarks anywhere for these specific cpu comparison.

Don't need the upgrade that much but I'd rather sell my 8120 and put a bit now to upgrade than having to shell out the whole retail price next year for another cpu when mine becomes dirt cheap in the second hand market.

If you look on the product pages for the PD cpu's OC have listed the % improvements.

Its not much, about 7% I think.

Martini1991 · 29 Oct 2012 at 11:36

opethdisciple said:
If you look on the product pages for the PD cpu's OC have listed the % improvements.

Its not much, about 7% I think.

In situations it's more like 7-15% tbh.

humbug · 29 Oct 2012 at 11:39

Martini1991 said:
1.) Isn't Overclockers.net

2.) No he didn't say that.

He did have a moan about what he could take the bus speed too (HT Reference clock) however, although I'm sure that's more down to the board than the chip.

Overclocked3D.net then

He moaned about the fact that he could not adjust the bus at all without it crashing, the whole video has nothing of any substance, its just 30 minuts of him whinging about the bus, it being a lot slower in one benchmark than Intel and one other thing that's already old news with this architecture.

He did use the time to say it ran a lot cooler than Bulldozer, about 30 seconds.

Martini1991 · 29 Oct 2012 at 11:49

He said he got it to 275.
I'd say that's an adjustment.

That said, he really must rant as I clicked in any random place and got that

humbug · 29 Oct 2012 at 11:50

Martini1991 said:
He said he got it to 275.
I'd say that's an adjustment.

That said, he really must rant as I clicked in any random place and got that

What time in?

passey89 · 29 Oct 2012 at 11:51

Davedree said:
passey I may be wrong but that the asrock 990fx extreme 3 only has a a 4+1 vrm for the power without a heatsink, so they might run hot/hold back you overclock with instability when you try and push the chip past its thermal limits. Ideally If i was investing in an am3+ back when they were coming out I would made sure it was 8+2. But i made a better choice and jumped ship and I don't regret it at all

It does only have the 4+1 but have heat sinked vrm's.

I've had 4.2 ghz out of my 1075t and its been perfectly fine @1.4V.

Thats a 33% overclock.

getting an 8350 to 5ghz will only be a 25% increase so as long as its around the same voltage should be fine.

Martini1991 · 29 Oct 2012 at 11:56

humbug said:
What time in?

Well there's the written review that accompanies it ;
http://www.overclock3d.net/reviews/cpu_mainboard/amd_vishera_fx8350_piledriver_review/2

Easier to clicking around like a lunatic to find it again.

Although I find it, he says it 13:12 in.

passey89 said:
It does only have the 4+1 but have heat sinked vrm's.

I've had 4.2 ghz out of my 1075t and its been perfectly fine @1.4V.

Thats a 33% overclock.

getting an 8350 to 5ghz will only be a 25% increase so as long as its around the same voltage should be fine.

That's not exactly impeccable logic

humbug · 29 Oct 2012 at 12:09

Yup found it ^^^^ its 2 seconds, i must have tuned out at that point with the constant ranting

@ passey89, they have different volts levels, your talking way over 1.4v

Hell i'm running 1.43v @ 4.2Ghz on Thuban, and that's good, i'm shocked to see you running 1.4.

4.7Ghz on Air - P-95; 1.48v, temp 62c

4.6Ghz on Air; 1.44v, temp 46c.

5Ghz on water; 1.54v temp 52c

Martini1991 · 29 Oct 2012 at 12:11

I wouldn't ever take an AMD temperature reading as evidence what it's running at.
I mean you only have to start thinking about the TDP at those levels of power consumption, and it's running cooler than things with a lower TDP?

Think you'd just have to go in blind, if it doesn't throttle and it's stable you're good to go

EDIT : I mean the evidence is staring you in the face, the ambient temperature is like 22-23c, and he's idling ~8c cooler than the ambient on air?

humbug · 29 Oct 2012 at 12:20

Martini1991 said:
I wouldn't ever take an AMD temperature reading as evidence what it's running at.
I mean you only have to start thinking about the TDP at those levels of power consumption, and it's running cooler than things with a lower TDP?

Think you'd just have to go in blind, if it doesn't throttle and it's stable you're good to go

Its irrelevant, AMD temps are read from 1# the silicone (cores) its self and 2# the underside of the socket.

So far Pilediver is stable at ~60c cores and 75c socket.

How accurate in relation to real life those temps may or may not be doesn't mater, its just stating something that has been true for AMD for 10 years, just in case people reading didn't know, eh?

Martini1991 · 29 Oct 2012 at 12:22

Of course it matters, and how's it irrelevant? You can't quote temperatures it's running at that it isn't actually running at.
It says it's running ~8c cooler than ambient idle.

Yeah sure, it's irrelevant.