Ok, second 8120 is bad...Is AMD miss-binningBulldozers?

R

Rauelius

2[H]4U
Joined
Jan 2, 2008
Messages
2,242
My question is, are the two defective FX -8120's I have recieve, cpu's that should have been binned as FX-6100 or FX-4100. The first one I had failed on Core 0, 3, 5. The one I got yesterday is failing on Core 7. If you have an 8120 I strongly recommend stress testing it at stock frequencies/voltages with OCCTP or AMD OverDrive. I could understand the first one being a fluke, but TWO in a row going bad, I smell somthing funky coming from Texas...I think AMD is mis-binning CPU's.

I got my RMA from newegg. OK, first thing I do whenever I get a new CPU is to stress it at stock, and make sure I have a good CPU before I go ahead and start overclocking it.

Well, first thing I do is run OCCTP, which has been extraordinarily useful for stressing CPU's. Well, two minutes into the test, it fails on "Core 07".

I thought to myself, "It's an old tool, maybe BD is weird with it"

So I downloaded the recently updated AMD OverDrive software, and used AMD's own software to test it. 5 minutes in, it fails.

Replaced the ram, with 8 GB of Corsair DDR3 1600 instead of the Avexir DDR3 1600, and it still fails on core 07 (Not that I thought RAM would have a big effect, but hey try anything).

So, right now, I upped the Voltage to 1.38 for the CPU and am testing it right now....let me see....yup still going...15 minutes strong!

Could this be a weird Vdroop issue? If this is successful, I'm gonna go for 4.2Ghz next with that same voltage.

Can anyone else with an 8120 try OCCPT or AMD OverDrive to test it. After getting two bad 8120's in a row, I feel that AMD is really blindly tossing the turds that couldn't quite run at 8150 speeds into the 8120 pile, when they really should be in the 6100 or 4100 pile.

The weird issues I'm experiencing at stock are very similar when ever I ran an unlocked Phenom II with no changes to voltage. For example, my Phenom II x3 740 unlocked to Quad Core would fail stress tests with that 4th core unlocked unless I upped the voltage.

Bear in mind, I am using a 900 Series board with the latest BIOS. Prior to installing the CPU I reset the BIOS and uninstalled AOD(amd overdrive). After getting these errors, I grabbed a spare HDD and reinstalled windows 7 on it and still received the errors. I did this with both defective 8120's to ensure there wasn't any weirdness from the previously installed 1090t.
 
what motherboard are you using? ive heard certain boards (gigabyte ud5) are known to have voltage drop issues. asrock 970 in your sig? I havent read about any issues with asrock 970. I had a 970 extreme that refused to work with my 1080w psu, so i sold it. Really to bad, i loved everything else about that board.
 
Try setting your stock voltage manually or just a tick higher then stock and testing it. If it still errors out then it's a bad cpu. Mobo makers take into account vdrop. Watch cpu-z when it's under load and see how much the voltage drops while testing as some vdrop is usually expected...
 
Upped it to 1.38v from the Stock 1.30 and it ran at stock fine. Tried OC it, and nothing but failure
 
Rauelius said:
Upped it to 1.38v from the Stock 1.30 and it ran at stock fine. Tried OC it, and nothing but failure
Click to expand...

My guess is they relaxed the QC binning process to get more cpu's out the door. You might never get an error under normal usage and a lot of users don't stress test their cpu's...unfortunately were hearing more of these kind of issues with BD...not cool AMD. :(
 
Yeah, with my two bad 8120's and the tons of other threads that I read on online sites, I really think AMD is binning failed 8150's as 8120's, when they should be FX 4100's (my first one had 3 bad cores) or FX 6100's (this current one has one bad core). It's not my PSU, I had my 1090t on it running at 4.2Ghz with two GTX460's @ GTX560 speeds. I even threw in my 8800GT in for PhysX (the AsRock is a cheap board, but it's really very nice!) and the machine is ROCK SOLID STABLE. When I reinstalled windows I pulled out my Trusty Radeon 5770 in single card config to see if it was a weird issue with nVidia cards...same problem...only running with the FX-8120.

Hell for the heck of it, I through on my Phenom II x3 740, unlocked it to quad and kept the voltages stock along with the clock speeds. Solid as a ROCK. This chip wasn't stable on stock Volts on my Asus 785 board but Rock solid on my ASRock 970.

Considering that I had TWO bad FX-8120's and the myriad of errors listed on this and many other boards, I really think AMD has to answer to this. I really don't want to think that AMD just threw random Bulldozers into a pile and binned them blindly only making sure the 8150's (which are as rare as hens teeth) worked and hoping that people who picked up the 8120 wouldn't test their chips.

This is an embarrasment, and I can't even RMA it for a 3rd shot because NewEgg is sold out. I was really looking forward to see what I can do with this chip. I had a TON of fun overclocking my 965 to 4.09Ghz and my 1090t.

I know the reviews were "meh" on BD, but hell...I got the Phenom 9600 all those years ago...and...it's still used in my homeserver.

I guess it works out. I paid $220 for this, and will consider trying it again after a price-correction to $170 and a few BIOS updates Maybe AMD will release an update that fixes the issues like they did with the Phenom I? The Phenom 9600 was replaced by the Phenom 9650, maybe AMD will make a Working FX-8125?
 
I've had an issue on my q6600 g0 stepping where core #3 will fail tests unless i up the voltage. The problem is /w my motherboard, and if i feed it enough juice its stable for eternity /w no failure. So i still think its your motherboard /w the new chip especially since its happend to you twice with two different chips....but i could be wrong....wouldnt be the first time
 
my 8150 doesn't pass AOD stress test stock settings, processor is under 35C too. Works fine if I manually set voltages and overclocks decent. default bios settings for Crosshair V, latest bios.
 
Rauelius said:
Then AMD must hate me.. :(
Click to expand...

Yeah, that sucks.

But I don't think AMD is misbinning 8120s. I know other people who've got better binned 8120s but I have an 8120 running at 5Ghz. Only stable through OCCT/Prime for around two hours then it craps out on me. Everything else is fine though. But if I really wanted to test stability, It's definitely rock solid at 4.88Ghz though 4 hours+. It's at 1.575v and I don't care about killing it because I'm upgrading to SB-E next month anyway.

 
Last edited:
It's a power hungry chip I bet it's voltage/power related, especially since you're on your 2nd one.
 
You know What, I'm playing around with the voltages as I'm typing this. CPU-Z reports vastly lower voltage than AMD OverDrive does. Also, Bios overclocking seems to be better. A shame, as I have grown fond of the tool. Ran this at 3850 speeds (Stock/Turbo/MaxTurbo) and ran fine with 1.38v.
Right now, I've got it at 4Ghz @ 1.48v...gonna test it. If it takes, should I cancel the RMA? Planned to run it 4.2Ghz with a MaxTurbo of 4.8Ghz.
 
Rauelius said:
You know What, I'm playing around with the voltages as I'm typing this. CPU-Z reports vastly lower voltage than AMD OverDrive does. Also, Bios overclocking seems to be better. A shame, as I have grown fond of the tool. Ran this at 3850 speeds (Stock/Turbo/MaxTurbo) and ran fine with 1.38v.
Right now, I've got it at 4Ghz @ 1.48v...gonna test it. If it takes, should I cancel the RMA? Planned to run it 4.2Ghz with a MaxTurbo of 4.8Ghz.
Click to expand...

I noticed the same thing, AMD overdrive reported higher voltage, once i installed software utility bundled with my motherboard it was showing me at 1.34V instead of 1.4 -- not only that but now i can see real cpu temps.
 
I cant blame you for being frustrated, with the one issue I was having Im aggrevated that the suggestion was an RMA on a backordered product. Its literally been the worst customer experience ever, I won't buy from them again after this. After I replied regarding the backorder with ' i'd had to be without a pc for potentially months to come after I payed for the product if rma'ing for a backordered item' I received a reply "It is up to you when you want to RMA the part. When you want to, you can submit an online warranty at http://support.amd.com/us/warranty/rma/Pages/rma-form.aspx" I mean yeah thats great, but how do you give that reply to someone that supported your not so great product launch? Not happy at all.

Anyone know any contacts at AMD to escalate an issue like this to?
 
Rauelius said:
You know What, I'm playing around with the voltages as I'm typing this. CPU-Z reports vastly lower voltage than AMD OverDrive does. Also, Bios overclocking seems to be better. A shame, as I have grown fond of the tool. Ran this at 3850 speeds (Stock/Turbo/MaxTurbo) and ran fine with 1.38v.
Right now, I've got it at 4Ghz @ 1.48v...gonna test it. If it takes, should I cancel the RMA? Planned to run it 4.2Ghz with a MaxTurbo of 4.8Ghz.
Click to expand...

I'd keep the chip for now...you have 3 years to RMA it correct, not sure what the warranty are on those things? I assumed you were using the bios for OC'ing, always use that for long term OC'ing. AOD can be funky at times, to many different motherboard configs/bios to deal with...
 
OK, I have it at 4.4Ghz, but It's strange. When ever I stress test it with either OCCTP or AOD it fluctuates and rarely hits 4.4Ghz, staying mostly at 3.8Ghz.However in SmallPT (which by the way has my FX-8120 @ 4.4Ghz scoring WORSE than my Phenom II 965 @ 4.09Ghz) all cores sit at 4.4Ghz, and I've seen it spike to 4.6 (first Turbo).
 
Ok 4.3Ghz @ 1.48v and all eight-cores are solid. Having it reliably Turbo up to 4.8Ghz with 1.55v and 4.6Ghz with 1.52v.

Is this a keeper?

I ran a test called SmallPt across my machines. Here are the scores:

2600k @ Stock = 80 Seconds
1090t @ 3.8Ghz = 163 Seconds
965BE @ 4.09Ghz = 203 Seconds
8120 @ 4.3Ghz = 212 Seconds

I know that BD is pretty crappy core vs core in comparison to the 2600k AND 1090t, but I expected in a multithreaded application like SmallPT I expected it to not only beat the 1090t, but annihilate the x4 965. Can anyone else try it and see if it's equally slow on their machines?
 
Last edited:
I contacted AMD about an issue with a thuban chip. took them 48-72 hours to respond -- they gave me 2-3 paragraphs and said yeah your welcome to RMA the chip and we will replace it. It left me with the impression I would be dead in the water for weeks. I much rather deal with newegg or other large e-tailers. With newegg, they dont have it listed on their website, but if you call, you can do an "advanced RMA" where they ship you the replacement upfront and you ship back them faulty product -- they only bill you if they dont receive faulty one after 14 days. I recently used this when i thought my 1080w PSU died (turned out to be the motherboard)

I explained two weeks is WAY to long for a computer nerd to be without a PSU.
 
buttons said:
I contacted AMD about an issue with a thuban chip. took them 48-72 hours to respond -- they gave me 2-3 paragraphs and said yeah your welcome to RMA the chip and we will replace it. It left me with the impression I would be dead in the water for weeks. I much rather deal with newegg or other large e-tailers. With newegg, they dont have it listed on their website, but if you call, you can do an "advanced RMA" where they ship you the replacement upfront and you ship back them faulty product -- they only bill you if they dont receive faulty one after 14 days. I recently used this when i thought my 1080w PSU died (turned out to be the motherboard)

I explained two weeks is WAY to long for a computer nerd to be without a PSU.
Click to expand...

Thank you sir, I'll go this route. Im debating waiting for the b3 to ship, its working to a degree right now and if I can avoid another b2 id rather. Really appreciate the heads up here, I wish the rep would have mentioend this via email He was rather short and rude with me and it turned me off towards AMD altogether tbh.
 
Rauelius said:
OK, I have it at 4.4Ghz, but It's strange. When ever I stress test it with either OCCTP or AOD it fluctuates and rarely hits 4.4Ghz, staying mostly at 3.8Ghz.However in SmallPT (which by the way has my FX-8120 @ 4.4Ghz scoring WORSE than my Phenom II 965 @ 4.09Ghz) all cores sit at 4.4Ghz, and I've seen it spike to 4.6 (first Turbo).
Click to expand...

The [H] review suggested cutting off Turbo when overclocking that high. Also I think it says something about OC'ing using AMD's OverDrive will cause the memory to run 25% slower. So you have to do it in the bios. Plus when it Turbo's I think it cuts off cores? I would read their review to get a basic setup to OC with. I think they were at 1.5v on the core in the end also.
 
OK, I have it at 4.4Ghz, but It's strange. When ever I stress test it with either OCCTP or AOD it fluctuates and rarely hits 4.4Ghz, staying mostly at 3.8Ghz.However in SmallPT (which by the way has my FX-8120 @ 4.4Ghz scoring WORSE than my Phenom II 965 @ 4.09Ghz) all cores sit at 4.4Ghz, and I've seen it spike to 4.6 (first Turbo).
Click to expand...

turn off turbo and disable cool and quiet as well as c1e

set your multiplier in bios to 4.4ghz. Load windows Open Amd Over Drive, open up turbo core, turn on, click ok, turn off click ok, close Amd Over Drive.

You will then run at 4.4ghz all the time, and it will not throttle.

My bulldozer is faster in all areas, even single threaded vs my old 1055T even when my 1055T was clocked to 3.8ghz. I take a screen shot of cinebench at 3.8ghz on the 1055.

bulldozer.jpg
 
I wish I had your luck ebduncan, I just tried firing up games...and it blue screens. That's with 3 different installations of Windows 7. Every game did this! Deus Ex, Dirt 3, SSFIV, Portal 1+2, Sims 3, everything crashes the machine. Even after passing 3 hours of OCCT, I tried 5 different sets of ram as well. With 2 bad 8120's in a row, I guess I have to wait for the B3 stepping.
 
Rauelius said:
Yeah, with my two bad 8120's and the tons of other threads that I read on online sites, I really think AMD is binning failed 8150's as 8120's, when they should be FX 4100's (my first one had 3 bad cores) or FX 6100's (this current one has one bad core). It's not my PSU, I had my 1090t on it running at 4.2Ghz with two GTX460's @ GTX560 speeds. I even threw in my 8800GT in for PhysX (the AsRock is a cheap board, but it's really very nice!) and the machine is ROCK SOLID STABLE. When I reinstalled windows I pulled out my Trusty Radeon 5770 in single card config to see if it was a weird issue with nVidia cards...same problem...only running with the FX-8120.

Hell for the heck of it, I through on my Phenom II x3 740, unlocked it to quad and kept the voltages stock along with the clock speeds. Solid as a ROCK. This chip wasn't stable on stock Volts on my Asus 785 board but Rock solid on my ASRock 970.

Considering that I had TWO bad FX-8120's and the myriad of errors listed on this and many other boards, I really think AMD has to answer to this. I really don't want to think that AMD just threw random Bulldozers into a pile and binned them blindly only making sure the 8150's (which are as rare as hens teeth) worked and hoping that people who picked up the 8120 wouldn't test their chips.

This is an embarrasment, and I can't even RMA it for a 3rd shot because NewEgg is sold out. I was really looking forward to see what I can do with this chip. I had a TON of fun overclocking my 965 to 4.09Ghz and my 1090t.

I know the reviews were "meh" on BD, but hell...I got the Phenom 9600 all those years ago...and...it's still used in my homeserver.

I guess it works out. I paid $220 for this, and will consider trying it again after a price-correction to $170 and a few BIOS updates Maybe AMD will release an update that fixes the issues like they did with the Phenom I? The Phenom 9600 was replaced by the Phenom 9650, maybe AMD will make a Working FX-8125?
Click to expand...

You do realize that by getting 2 bad CPU's in a row the odds are astronomical against it happening! Your problem may just be other hardware! BullDozer draws alot of current more than any Phenom2 does even at stock speeds for the BD.So when you install another cpu of an earlier AMD chip it's not going to give apples to apples results no matter how you overclock it. Slow down and go through everything slowly including reloading your OS Everytime you change a component. Remember bluescreens can be for a variety of reasons hardware wise ,ram -ram voltage, CPU cpu voltage,harddrive errors,mobo not carrying enough current ,psu not supplying enough current. Don't trust anything but a digital volt meter to check your voltages; bios or software is notorious for errors in reporting incorrect voltages Good Luck !! Hope it is something other than your cpu!! One other thought be sure and turn off any Spread Spectrum entries in the bios along with cool and quiet. These things have caused me problems in the past!! You do have the latest bios for this board ??
 
Rauelius said:
I wish I had your luck ebduncan, I just tried firing up games...and it blue screens. That's with 3 different installations of Windows 7. Every game did this! Deus Ex, Dirt 3, SSFIV, Portal 1+2, Sims 3, everything crashes the machine. Even after passing 3 hours of OCCT, I tried 5 different sets of ram as well. With 2 bad 8120's in a row, I guess I have to wait for the B3 stepping.
Click to expand...

My guess is this has to be power related. The reason being is running games vs windows, thats the only difference is a lot of extra juice coming from your videocards for graphics. Once your videocards kick in, not enough juice, bluescreen.

Hammering the cores all day wont tell you anything about your rig when it comes to stable gaming. You would have to load the GPU's.

Have you tried furmark? I would bet if you fire up furmark, it bluescreens as well. You should try a different power supply just for shits and giggles if you can. Just to see if it makes a difference. We have all read that BD takes a lot more juice then P-IIX4, that along with your vidcards could be too much for your current PWS. Just my guess anyway. Good luck.

EDIT: If you dont have another PWS thats big enough, you can always paperclip another PWS and run your graphics only from the second PWS and keep your current one in, this will take the load of your cards off your current one and see if you can then game without bluescreening. That should also work, assuming this other PWS you use has enough juice to run your graphics. I know you mentioned crossfire in there somewhere in a previous post.
 
Frank_G said:
Slow down and go through everything slowly including reloading your OS Everytime you change a component. Remember bluescreens can be for a variety of reasons hardware wise ,ram -ram voltage, CPU cpu voltage,harddrive errors,mobo not carrying enough current ,psu not supplying enough current. Don't trust anything but a digital volt meter to check your voltages; bios or software is notorious for errors in reporting incorrect voltages Good Luck !! Hope it is something other than your cpu!! One other thought be sure and turn off any Spread Spectrum entries in the bios along with cool and quiet. These things have caused me problems in the past!! You do have the latest bios for this board ??
Click to expand...
Are you high?
Reload windows every time you replace ram? and cpu? eh? should i reload the os every time i change my overclock too?
where exactly do i measure the vcore with a volt meter?

hell i am running the same os without a reinstall from a p4 3.0ghz, c2d e6320, to my current Athlon II x2 260 @4.1ghz.
 
I've been experimenting with BD on my AMD rig.

In lightly threaded games it seems like I get stutter in games with turbo core on. It seems as if windows is bouncing the load around and cores are going up and down in speed.

When I disabled turbo core and clocked up where I wanted to be stutter disappeared. The games were smooth and then I started getting random crashing. Ugh, headache.

As far as Cine goes my old 1055T @ 4.1GHz got 7.29.

 
I had turbo turned off in the BIOS and AOD disabled in the CCC. Now I've killed low power states as well. It seems as if games are now smooth with no stutter or crashing. I'll update if that changes.
 
Many SteamWorks games are not compatible with Bulldozer. Deus Ex, Dawn of War, Portal 2, Total War Shogun 2, etc will crash you PC 100% of the time. There is supposed to be a fix in the works, but I've never read where AMD confirmed the issue officially.
 
Blacklash said:
I had turbo turned off in the BIOS and AOD disabled in the CCC. Now I've killed low power states as well. It seems as if games are now smooth with no stutter or crashing. I'll update if that changes.
Click to expand...

Turbo disables cores to increase clock speed. Here's a quote from the [H}] review.


Words about AMD Turbo Scaling & Power Savings Features
We have run all of our stock processor clocks with Turbo turned on and power saving features enabled. On our overclocked processor tests, we have turned off these features. We think this is how you are most likely to use the processors in a real world situation.

As you can see on the charts above, AMD FX processors have three clock speeds listed. These speeds outline the stock clock and Turbo clocks the processors use.

We are using the Corsair H100 for all testing here today. Certainly, a different cooling solution could change possibly change the scaling you would see at home as the processor temperatures ramp up.

AMD FX-8150 Turbo Scaling - 8/7 thread load = 3.6GHz - 6/5 thread load = dynamic scaling between 3.6GHz and 3.9GHz - 4 thread load = 3.9GHz - 3 thread load = dynamic scaling between 3.9GHz and 4.2GHz - 2/1 thread load = 4.2GHz.

AMD FX-8120 Turbo Scaling - 8/7 thread load = 3.1GHz - 6/5 thread load = dynamic scaling between 3.1GHz and 3.4GHz - 4 thread load = 3.4GHz - 3 thread load = dynamic scaling between 3.4GHz and 4GHz - 2/1 thread load = 4GHz.


I would imagine that if you're near the end of your overclocking potential and then expect the chip to "Turbo" for another 600 MHz you're going to have stability problems. Mine when it "Turbo'd" it took my voltage from 1.4125v to over 1.6v. If you have the cooling then sure that might be fine, but for most of us it's going to fail the system.

Stability > Tricks when OC'ing.
 
No. I do not manually OC with turbo enabled.

When I was first testing BD it was at default settings without manual overclocking. That's when I noticed the occasional stutter in some games.

When I began manually overclocking I disabled turbo and AOD. That's when I started seeing some games occasionally crashing but there was no stutter while playing. When I disabled low power states that problem vanished as well. I should note my manual OC was fine when tested with OCCT. The problem only appeared in some games with low power states enabled.
 
Last edited:
You must log in or register to reply here.
Back
Top