AMD FX OC'ers Club

[Alcpone]

Cooler master T2. Got a fairly hefty surface area on the ali and a 12cm fan. I'm going to put the stock fan off the billy basic cooler above the gets as I think keeping them cool is going to be my biggest headache. But if I can get 4.4ghz ish I will be happy. Maybe I will up the nb and maybe get a higher clock out the gpu? Will have a play.

[Irony]

Oh ok. I think 4.4 should be doable.

[Alcpone]

How did you manage getting 4.5Ghz on stock volts? Hopefully my cooler will be here today so I can get started clocking on the weekend

[Irony]

When I first started it seemed kinda volt hungry, needed 1.4 or so for that. but after a couple weeks I redid my oveclock going in really small increments and I made it all the way to 4.5 on stock volts. Somehow it seemed like it got more stable. 4.6 takes about 1.4 I think, and 4.7 was almost 1.5, then anything above that I have had lots of trouble getting stable.

Also, arctic silver5 takes way too bloody long to cure. It was several weeks before I was getting acceptable temps.

[Batou1986]

I still see ppl using prime95 for stability testing
Prime is not suitable for modern cpu's it will not stress them enough to find faults or reach max temp

[Irony]

Ya and on these piledrivers prime says that cores are failing no matter what

[Alcpone]

Well got the monster on. Had to take the case fan and put it on the back instead, no big deal just means I can direct the cool air over the mosfets
Waiting on the sinks coming from china so not going to push it until they come but even at 4.1 it is way cooler. Will I notice better temps when the paste settles? I put just enough on to cover without squishing out the sides.

[theoneandonlymrk]

Originally Posted by eidairaman1

its interesting that voltage deltas are pretty much the same as the Athlon XP days

Indeed , bit weird.
Another thing ive noticed is that regardless of applied coreV (1.45v on mine @4.74ghz) the cpuidv in overdrive never changes ie the per core applied core volts(1.332).
This can be upped only by the overdrive utility.
What's going on here? Is it a software misread, I havent had the time to investigate it

[n0tiert]

i´ll make some test FX-8350 vs. FX-8150
as soon the new build is done..... they still charge you 50+ Euro more on the FX-8150

[theoneandonlymrk]

Originally Posted by n0tiert

i´ll make some test FX-8350 vs. FX-8150
as soon the new build is done..... they still charge you 50+ Euro more on the FX-8150

http://img.techpowerup.org/130223/P1010030.jpg

Very interested in your 8350 oc results as ive a T chip and you have a N plus yours is a different batch

[os2wiz]

Originally Posted by theoneandonlymrk

Very interested in your 8350 oc results as ive a T chip and you have a N plus yours is a different batch

How do you tell N vs T and what significance does it have. Also his batch number is unusual . It is 1302. I guess after they run 01-12 on first two digits they continue upwards so that you dont mistake a December 2012 chip with a December 2013 chip. So 13 is January 2013 and a December 2013 chip would be 24--.

[theoneandonlymrk]

Originally Posted by os2wiz

How do you tell N vs T and what significance does it have. Also his batch number is unusual . It is 1302. I guess after they run 01-12 on first two digits they continue upwards so that you dont mistake a December 2012 chip with a December 2013 chip. So 13 is January 2013 and a December 2013 chip would be 24--.

I don't know what significance it has but you can see on his pic a little N on the corner of the interposer top right(interface pcb) not on the ihs mine has a T, cadaveca hinted at possible differences but couldn't elaborate which has left me intrigued

[os2wiz]

Originally Posted by theoneandonlymrk

I don't know what significance it has but you can see on his pic a little N on the corner of the interposer top right(interface pcb) not on the ihs mine has a T, cadaveca hinted at possible differences but couldn't elaborate which has left me intrigued

Yes I can see it when I zoom the picture. Cadaveca did not indicate whether T was better than N did he???

[theoneandonlymrk]

Originally Posted by os2wiz

Yes I can see it when I zoom the picture. Cadaveca did not indicate whether T was better than N did he???

No , that just makes it more intriguing

[os2wiz]

Originally Posted by theoneandonlymrk

No , that just makes it more intriguing

I do not like man-created mysteries. Let the light shine on it.

[cadaveca]

Originally Posted by os2wiz

I do not like man-created mysteries. Let the light shine on it.

The answer is that I don't know, but I think it's something we should pay attention to. I nthe past, I've found "T" chips to be better, and be able to handle higher voltage. But it's been some time since I've had more than a few AMD CPUs of the same model to play with. Back in 939 days I must have had over 1000 chips go through my hands, and I do think "T" chips seemed better then.

That might not apply now, however.

The only way to find out is to track it, I guess. It might prove unimportant.

[os2wiz]

Originally Posted by cadaveca

The answer is that I don't know, but I think it's something we should pay attention to. I nthe past, I've found "T" chips to be better, and be able to handle higher voltage. But it's been some time since I've had more than a few AMD CPUs of the same model to play with. Back in 939 days I must have had over 1000 chips go through my hands, and I do think "T" chips seemed better then.

That might not apply now, however.

The only way to find out is to track it, I guess. It might prove unimportant.

What percentage of the FX chips are T versus n? If it is a very low percentage it may prove difficult to track and database any trend. Few of us have access to enough chips to run into a T chip. And too few of us would even know to look for it so we could keep track.

[theoneandonlymrk]

Originally Posted by os2wiz

What percentage of the FX chips are T versus n? If it is a very low percentage it may prove difficult to track and database any trend. Few of us have access to enough chips to run into a T chip. And too few of us would even know to look for it so we could keep track.

We can make a start

[cdawall]

Originally Posted by cadaveca

The answer is that I don't know, but I think it's something we should pay attention to. I nthe past, I've found "T" chips to be better, and be able to handle higher voltage. But it's been some time since I've had more than a few AMD CPUs of the same model to play with. Back in 939 days I must have had over 1000 chips go through my hands, and I do think "T" chips seemed better then.

That might not apply now, however.

The only way to find out is to track it, I guess. It might prove unimportant.

Location on the wafer makes more of a difference than T vs N as well as the batch of the individual chips.

[theoneandonlymrk]

Originally Posted by cdawall

Location on the wafer makes more of a difference than T vs N as well as the batch of the individual chips.

Whilst I don't disgree in a way I do as defects dont happen in any particular area .
Anyway a statement like that should only be made if you KNOW what the T or N mean. ... do you. ?

[os2wiz]

Originally Posted by cdawall

Location on the wafer makes more of a difference than T vs N as well as the batch of the individual chips.

I suspected that location on the wafer was important as you see so many differences in over clocking even in the same batch number with proper motherboard selection. I have batch 1236 8350 CPU that is running on the Asus Cross hair V. (Not the Formula Z) . I can only over clock to 4.78 GHZ and that is at 1.52 volts. At 4.6 GHZ I only require 1.41 volts. VID is 1.325 v. I have a H100 liquid cooler. My temps go sky high at 4.78 GHZ
Under stress like. 67 Celcius with OCCT after 4 or 5 minutes that is core temp. I know the H100 is usually not good enough for 5 GHZ but for 4.8 it should be adequate. I am opting for a Swiftech H320 Liquid cooler that has a 360 mm copper radiator better tubing and better pump than Corsairs junk. It should be to market in about 6 weeks. Then I'll be able to push the CPU a little further. But it is obvious I have one of the poorer chips from my batch.

[n0tiert]

the RIG is up´n running, need to rush @ work.... will show some FX-8350 vs. FX-8150 tonight...... maybe we get behind that "N"

[Alcpone]

So I am currently sitting at 4.1Ghz before I push on, once my heatsinks for the mosfets arrive. I have angled airflow over them so that should help keep them cool, the CPU is at 1.275v to get 4.1Ghz stable and sitting at 47/48c at 100% load. So my question do you think I have a eco chip that needs very little power to get high clock's? So in theory I should beable to get a high OC? Maybe my board is going to be the weakest link? I have run a few benchmarks and I am quite suprised what it actually runs fairly parallel with other more powerful chips.

[n0tiert]

Chip AMD FX-8150




3DMark AMD FX-8150







Cinebench 11.5 AMD FX-8150





Unigine 4.0 DX11 AMD FX-8150





all Settings are Stock !!!
more to come (8350 vs. 8150) or gimme some info what you like to see before i´ll switch the CPU (benches)

[cdawall]

Originally Posted by theoneandonlymrk

Whilst I don't disgree in a way I do as defects dont happen in any particular area .
Anyway a statement like that should only be made if you KNOW what the T or N mean. ... do you. ?

I have discussed it on XS with others and there are some ideas floating around. No one outside of AMD seems to really know for sure what it references to. The letters do not line up with manufacturing plants, processes or even substrate materials. It might be that the letter refers to location on the wafer itself which would explain a lot about why there are substantially more of one chip than the other. There is a higher likelihood an edge chip would be defective and there are less of them to begin with.

Now why since the dawn of 754/939 this has worked out and been documented that T chips are high voltage CPU's and N chips are low voltage CPU's I do not know. In all honesty it could purely be AMD binning CPU's at the wafer level.