- Joined
- Oct 14, 2007
Hey TJ, here's a new tool for you: http://hwbot.org/forum/showthread.php?t=59243
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
TJ rips into a New APU - 5800k on ga f2a85x-up4
- Thread starter TsunamiJuan
- Start date
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #22
You rock man, thanks for that 
. Stilt was a huge help last time to actually hes got a bunch of great info.
. Stilt was a huge help last time to actually hes got a bunch of great info.- Joined
- Jun 18, 2012
- Location
- Stark Industries
Thanks TJ, and keep the fm2 stuff coming!
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #24
I am gonna Rip into this more today with the new utility. I am tempted at this point to just disk bricking another board since the Release bios is crap, and i can't seem to get my hands on something besides f3g. I do have a feeling that I will probably end up abandoning this board if the other fm2 board i have on the way is really as good as it looks. It certainly looks good on paper.
Gonna be giving stilts tools a shot today I am pretty sure by adjusting the limiter values I can nudge a stable 4.8ghz out of it on water while doing 3d @ close to 1000mhz igp speed.
Gonna be giving stilts tools a shot today I am pretty sure by adjusting the limiter values I can nudge a stable 4.8ghz out of it on water while doing 3d @ close to 1000mhz igp speed.
- Joined
- Jun 18, 2012
- Location
- Stark Industries
You mean risk..man u need a secretary to take those calls
caller: Hi Juan there?
secretary: Yes. But he aint talking to u.
caller: y not?
secretary: He's busy typing on overclockers forum. and spell checking.
caller: WTF!
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #26
what a weekend, what a weekend.
So I tried to make a post last night, but I ended up forgetting to hit submit. Since I got side tracked cause I was benching while writing it up.
Got some nice results over the weekend to many to really name, there are some that speak above the others. In truth none are exceptional but the overall Efficiency of the submissions does speak well for the platform. I didn't end up killing another board which is also a positive. There was definitely something else wrong with that other board. I only ended up having recover the bios twice on the new board. Due to excessive memory corruption. FUN FUN FUN
My limitation this weekend was really my ram and timings. While cas 10 RipsawZ's are good ram. They are so high strung at the default values that while pushing this high of a Northbridge frequency there just wasn't much room for error. This reared its ugly head alot during the weekend. Between water and phase change. I did figure out how to get around it somewhat. But I really need to set down and relearn memory on this platform. For some reason this bios Hates my evo's which are a similar speed to the Ripsaws. Though the ripsaws are faster. I have a set of corsair Dominator Platinum's coming sometime mid week, that are rated for 2666, which should have room to push to 2800+. I was gonna buy a set of 2800 sticks so that I had some with SPD's already setup and factory rated for such work. For use with reviews and general Shennanigans. But They where a little out of my budget this week. Leave it to someone like me to spend the amount of money most people spend on a real system on a value setup. Though what I have bought is really for benching in general so its not limited to one system.
So with the limitations due to ram I saw some wonderful Clock speeds both core and igp. On the hole I ran the bulk of the weekend with the ram at cas10 @2400mhz with a nb frequency of 2500mhz. With a little playing around over the week i should be able to figure out how to get this to be more stable. This helps the IGP performance out alot. I managed to push a wonderful 1268mhz out of the IGP in just about every benchmark. I know theres room for it to push further Since i ran a few at 1300-1370mhz. However I didn't end up completing the benchmarks due to a mix of Heat and sub timing issues for the bus. Should be able to iron this out. I don't see any reason why the IGP on these chips wouldnt be capable of running at 1350-1400mhz On air/water if you have a good setup. You wouldn't be running a high core clock speed. But you could certainly Run an extremely high IGP speed and have plenty of room in the thermal envelope.
Core speeds for water where around 4.8-4.9ghz Maximum. Even then that was kinda pushing it and I was starting to see diminishing returns from the heat. Still much better than anything you could of hoped for on the Llanos. Phase change performance Seems to be pinned at around 5.1-5.2 ghz Atleast on my small system. Though Honestly the heat output on these chips is much much lower than once again compared to the llanos. On the hole I only saw 8c in temperature swing on the phase change unit. During a 13 Hour run. Usually I end up poisioning the system with excessive heat and the core ends up into the positives after a while. Not the case with the a10's. I sat at -32c to -40c All day long.
On the hole this all speaks wonders for the platform. It also makes me wonder is it even worth it to pour ln2 on this setup. (wow yeah i really did say that) By the current platform benchmarks out there. If I fixed my memory problems I am already close to what the guys running ln2 are doing. Extremely Close, in fact its really the difference in ram speed thats holding me back. Not so much the clocks I am achieving. Though I will be able to push higher clocks when the ram timings get fixed. Look for yourself's though at the hwbot rankings currently for the platform, you'll see my train of though. Some of this probably has to do with the fact that the people playing with trinity are not big APU overclockers. They are mostly reviewers putting down base numbers. Similar to what I did over the weekend. It will be interesting to see how this changes over the next few months.
So I tried to make a post last night, but I ended up forgetting to hit submit. Since I got side tracked cause I was benching while writing it up.
Got some nice results over the weekend to many to really name, there are some that speak above the others. In truth none are exceptional but the overall Efficiency of the submissions does speak well for the platform. I didn't end up killing another board which is also a positive. There was definitely something else wrong with that other board. I only ended up having recover the bios twice on the new board. Due to excessive memory corruption. FUN FUN FUN
My limitation this weekend was really my ram and timings. While cas 10 RipsawZ's are good ram. They are so high strung at the default values that while pushing this high of a Northbridge frequency there just wasn't much room for error. This reared its ugly head alot during the weekend. Between water and phase change. I did figure out how to get around it somewhat. But I really need to set down and relearn memory on this platform. For some reason this bios Hates my evo's which are a similar speed to the Ripsaws. Though the ripsaws are faster. I have a set of corsair Dominator Platinum's coming sometime mid week, that are rated for 2666, which should have room to push to 2800+. I was gonna buy a set of 2800 sticks so that I had some with SPD's already setup and factory rated for such work. For use with reviews and general Shennanigans. But They where a little out of my budget this week. Leave it to someone like me to spend the amount of money most people spend on a real system on a value setup. Though what I have bought is really for benching in general so its not limited to one system.
So with the limitations due to ram I saw some wonderful Clock speeds both core and igp. On the hole I ran the bulk of the weekend with the ram at cas10 @2400mhz with a nb frequency of 2500mhz. With a little playing around over the week i should be able to figure out how to get this to be more stable. This helps the IGP performance out alot. I managed to push a wonderful 1268mhz out of the IGP in just about every benchmark. I know theres room for it to push further Since i ran a few at 1300-1370mhz. However I didn't end up completing the benchmarks due to a mix of Heat and sub timing issues for the bus. Should be able to iron this out. I don't see any reason why the IGP on these chips wouldnt be capable of running at 1350-1400mhz On air/water if you have a good setup. You wouldn't be running a high core clock speed. But you could certainly Run an extremely high IGP speed and have plenty of room in the thermal envelope.
Core speeds for water where around 4.8-4.9ghz Maximum. Even then that was kinda pushing it and I was starting to see diminishing returns from the heat. Still much better than anything you could of hoped for on the Llanos. Phase change performance Seems to be pinned at around 5.1-5.2 ghz Atleast on my small system. Though Honestly the heat output on these chips is much much lower than once again compared to the llanos. On the hole I only saw 8c in temperature swing on the phase change unit. During a 13 Hour run. Usually I end up poisioning the system with excessive heat and the core ends up into the positives after a while. Not the case with the a10's. I sat at -32c to -40c All day long.
On the hole this all speaks wonders for the platform. It also makes me wonder is it even worth it to pour ln2 on this setup. (wow yeah i really did say that) By the current platform benchmarks out there. If I fixed my memory problems I am already close to what the guys running ln2 are doing. Extremely Close, in fact its really the difference in ram speed thats holding me back. Not so much the clocks I am achieving. Though I will be able to push higher clocks when the ram timings get fixed. Look for yourself's though at the hwbot rankings currently for the platform, you'll see my train of though. Some of this probably has to do with the fact that the people playing with trinity are not big APU overclockers. They are mostly reviewers putting down base numbers. Similar to what I did over the weekend. It will be interesting to see how this changes over the next few months.
- Joined
- Jun 18, 2012
- Location
- Stark Industries
Thanks for the update.
TJ, any reason y u r going for high igp clocks?
I may have ask u this b4 - i notice on llano higher igp clocks did not render more fps in actual gaming.
TJ, any reason y u r going for high igp clocks?
I may have ask u this b4 - i notice on llano higher igp clocks did not render more fps in actual gaming.
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #28
Because not only did i want to verify what they claimed the baselines where. I wanted to see how much of an improvement was made in the overall chip design.
The power management on the llanos Tends to prohibit high IGP clocks at high clock speeds. You just generally can't get enough power into the chip and adjust things right that it doesnt protect itself. So things like air/water and phasechange cooling are the best places to test IGP for the llano.
Also I buy these chips for one reason only. I enjoy benching them. I enjoy pushing what they can do in benchmarks. I Like that AMD chips Scale like crazy when cold, and I like that they are a serious challenge to work with. Compared to Sandy Bridge and some of intels more current stuff. It generally Takes Patience, Time, and extended knowledge to really push them.
When it comes to real world application, I admit I don't use one everyday. I Could probably easily do the bulk of what I want to do one one of these. In fact I will probably end up Setting a system up to see how well it fares for everday use at one point. I easily found that a decent amount of the gaming I did could be done on the llanos. There where a few pieces of software out there that didn't like them. Primarily anything that had a high multicore load 3+ cores in use. Due to the power management. We'll see as time goes on and i play with the platform more if the A10's can reign in the rest of the applications the llanos where incapable of playing smooth enough for someone as picky as I am. Though one big level of impressive performance on the llanos was that I could actually Play World of Warcraft In a 25man raid, at max detail. Granted the system was overclocked. Nothing serious only 4.2 ghz' or so. But It was a good experience. Which speaks well for the market itself since alot of MMO gamers don't generally have the money for super high end machines.
The power management on the llanos Tends to prohibit high IGP clocks at high clock speeds. You just generally can't get enough power into the chip and adjust things right that it doesnt protect itself. So things like air/water and phasechange cooling are the best places to test IGP for the llano.
Also I buy these chips for one reason only. I enjoy benching them. I enjoy pushing what they can do in benchmarks. I Like that AMD chips Scale like crazy when cold, and I like that they are a serious challenge to work with. Compared to Sandy Bridge and some of intels more current stuff. It generally Takes Patience, Time, and extended knowledge to really push them.
When it comes to real world application, I admit I don't use one everyday. I Could probably easily do the bulk of what I want to do one one of these. In fact I will probably end up Setting a system up to see how well it fares for everday use at one point. I easily found that a decent amount of the gaming I did could be done on the llanos. There where a few pieces of software out there that didn't like them. Primarily anything that had a high multicore load 3+ cores in use. Due to the power management. We'll see as time goes on and i play with the platform more if the A10's can reign in the rest of the applications the llanos where incapable of playing smooth enough for someone as picky as I am. Though one big level of impressive performance on the llanos was that I could actually Play World of Warcraft In a 25man raid, at max detail. Granted the system was overclocked. Nothing serious only 4.2 ghz' or so. But It was a good experience. Which speaks well for the market itself since alot of MMO gamers don't generally have the money for super high end machines.
watchman323
New Member
- Joined
- Oct 30, 2012
TsunamiJuan,
I am thinking about getting this motherboard with A10-5800K chip and 2400 ram as well . Do you think that you can publish your result on here, like the voltages for cpu, memory, and other setting that you set in the BIOS to over clock this baby?
Thanks
I am thinking about getting this motherboard with A10-5800K chip and 2400 ram as well . Do you think that you can publish your result on here, like the voltages for cpu, memory, and other setting that you set in the BIOS to over clock this baby?
Thanks
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #30
If your willing to wait about a week or so for results yes I will be posting a bunch across the spectrum, due to writing up some reviews on the platform. The voltages vary so much across the chip you have that my general voltages will not help you any. Not to mention I am on the high end of them since I am using more extreme methods of cooling. Which results normally in the board refusing to post and possibly bricking a bios.
- Joined
- Oct 30, 2012
So glad I found this thread. I just bought the A10, along with the ASRock FM2A85x and some GSkill 2400. Came here to figure out what I could run the NB voltage at to be stable at 2400 (2133 works at stock voltages, with 1.65v to the RAM). Looking forward to your results!
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #32
You shouldnt have to push the NB at all to run 2400mhz ram. If your not running the latest bios do that first. Thats probably whats holding you back. The release bios has memory compatibility issues on every board. As rock already has 3 bioses released for that board. So its probably been solved already.
The voltages on the fm2 generally auto adjust themselves pretty much for stock. Its already set on a per chip basis depending on how they binned coming out of the factory. This stays true till you really start pushing the chip while overclocking. However if your just trying to force the chip to run at 42x all the time it shouldnt really need much help in the voltage department at all.
The voltages on the fm2 generally auto adjust themselves pretty much for stock. Its already set on a per chip basis depending on how they binned coming out of the factory. This stays true till you really start pushing the chip while overclocking. However if your just trying to force the chip to run at 42x all the time it shouldnt really need much help in the voltage department at all.
- Joined
- Oct 30, 2012
Wow, thanks for the fast response, I'm honored! I didn't see the available BIOS updates my first few visits to the ASRock site, but I found them now. Cool, now I can share 2GB for video too, perfect! Thanks so much for the advice. I'll try it right now and POST results (ha)
Before I go, I'm trying something a little different with this build: I'm pretty satisfied with the CPU's performance (coming from a A64 X2 3600+), so I'm going to stick with stock CPU clocks (Turbo Boost to 4.2) in order to keep the Cool 'n' Quiet active. I kinda like the idea of running the CPU at 1.4GHz and 0.9V when it can (20W TDP!), and a manual overclock would disable that feature.
What I'm really after is running DDR3-2400 (at 1.65V) and clocking the GPU to about 1.1GHz (from 800). I want to see how much graphics performance I can get out of this thing. SO psyched to see a BIOS update this soon that opens the share to 2GB! Nice!
Before I go, I'm trying something a little different with this build: I'm pretty satisfied with the CPU's performance (coming from a A64 X2 3600+), so I'm going to stick with stock CPU clocks (Turbo Boost to 4.2) in order to keep the Cool 'n' Quiet active. I kinda like the idea of running the CPU at 1.4GHz and 0.9V when it can (20W TDP!), and a manual overclock would disable that feature.
What I'm really after is running DDR3-2400 (at 1.65V) and clocking the GPU to about 1.1GHz (from 800). I want to see how much graphics performance I can get out of this thing. SO psyched to see a BIOS update this soon that opens the share to 2GB! Nice!
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #34
You can actually leave clean and quiet on with fm2's and still overclock quite a bit, with the way the power profiles are set for these chips. You specify the low and max speeds for the chip. with clean and quiet off it will still cycle some between the set levels of boost and none boost so if you have it set at a 44x and boost a 46x it will go between those two settings based on load. Or you can leave clean and quiet on and it will auto choose the multis based on load so if your idle it will drop down to 800 or 1600ish at 1.2v and then depending on the number of cores in use from there and temp it will cycle up to the higher multis.
- Joined
- Oct 30, 2012
Thanks again, you rock! Well, I promised you results, and here they are:
After BIOS update, rebooted to Windows. No problems (kept previous 2133 clock).
Reboot and enter BIOS, set GPU memory window to 2GB (flippin' sweet), set RAM to 2400, all settings Auto.
Booted into Windows (Damn fast I might add; 6 seconds to logon screen after POST?!)
Logon, got desktop, icons, taskbar, mouse pointer, programs and services starting up, moved the mouse around for a few seconds, clicked the taskbar...
Froze up. So close. I mean, it probably would've frozen the second I played a game, but still encouraging progress.
Tried setting manual timings, being very generous, and that just made it worse (no Windows load screen at all). Uninstalled the ASRock XTreme Tuner Utility (thought that might be tweaking things behind my back), still wouldn't get there.
I'm definitely going to play around with the NB voltage after work tomorrow and get it stable at the desktop. The really good news is that with the previous BIOS NB voltage set at 1.2V, I got no further than the logon screen, and this time I made it to the desktop at Auto setting (1.75V, IIRC).
Thanks again for pointing out the BIOS, you've been the most helpful person out there so far! This platform is so new, I guess I'm one of the guinea pigs now. It's a cool feeling. I'll give you all the voltages/timings tomorrow when I'm replying at DDR-2400!
After BIOS update, rebooted to Windows. No problems (kept previous 2133 clock).
Reboot and enter BIOS, set GPU memory window to 2GB (flippin' sweet), set RAM to 2400, all settings Auto.
Booted into Windows (Damn fast I might add; 6 seconds to logon screen after POST?!)
Logon, got desktop, icons, taskbar, mouse pointer, programs and services starting up, moved the mouse around for a few seconds, clicked the taskbar...
Froze up. So close. I mean, it probably would've frozen the second I played a game, but still encouraging progress.
Tried setting manual timings, being very generous, and that just made it worse (no Windows load screen at all). Uninstalled the ASRock XTreme Tuner Utility (thought that might be tweaking things behind my back), still wouldn't get there.
I'm definitely going to play around with the NB voltage after work tomorrow and get it stable at the desktop. The really good news is that with the previous BIOS NB voltage set at 1.2V, I got no further than the logon screen, and this time I made it to the desktop at Auto setting (1.75V, IIRC).
Thanks again for pointing out the BIOS, you've been the most helpful person out there so far! This platform is so new, I guess I'm one of the guinea pigs now. It's a cool feeling. I'll give you all the voltages/timings tomorrow when I'm replying at DDR-2400!
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #36
Auto for ram on the platform is still pretty buggy. Your better off using the XMP to read your memory sticks profiles. I am not sure if the AsRock boards support this feature. I know the Gigabyte boards added it recently. I would assume that people would be quick to support this since its allways been kinda a sore spot for amd's release bioses with APU's. I had a similar problem when trying to get ram to get detected correctly which resulted in instability like your describing.
- Joined
- Jun 18, 2012
- Location
- Stark Industries
Hi vGorilla, am I reading this wrong? NB 1.75v? R u on extreme cooling as well?
On Llanos everydayOC Im on 1.2-1.3v closed water loop. Or FM2 NB is just that different?
- Joined
- Oct 30, 2012
Oops, nope, that should read 1.175, not 1.75. God no! I had it up to 1.2 the other day, but that was with the old BIOS. I'm just about to play around with it, I really want to get it stable at 2400, I'll post here in a half hour (hopefully) with loads of voltages and timings for y'all.
And TJ, yes, the board supports Intel XMP, I'll be sure to use it instead of Auto. I had thought it was a sort of memory standard, kinda like JEDEC, that prescribes various timings for various clock rates, but you say it actually reads/adjusts stuff? I missed a lot since the Athlon 64 days lol! And I'm really trying to not jack your thread, so my next post will definitely have some results for you. Thanks for the help so far.
And TJ, yes, the board supports Intel XMP, I'll be sure to use it instead of Auto. I had thought it was a sort of memory standard, kinda like JEDEC, that prescribes various timings for various clock rates, but you say it actually reads/adjusts stuff? I missed a lot since the Athlon 64 days lol! And I'm really trying to not jack your thread, so my next post will definitely have some results for you. Thanks for the help so far.
Last edited:
- Joined
- Oct 30, 2012
Running at 2400 now! Here are the current settings:
CPU Voltage - 1.35000
Boost speed - x42 4200MHz
CPU speed - x38 3800MHz
CPU NB voltage - 1.28125 (raised from 1.175 stock)
XMP Memory Profile 1.3
DDR3-2400
10-12-12-31
1.65V
Temperatures in BIOS -
At boot - 24.0 C MB, 27.0 C CPU
After 2 mins - 26.0 C MB, 32.5 C CPU
Well, I typed this up in Notepad; the previous vNB of 1.275 crashed after 5 minutes of Firefoxing. I'm certain that the current vNB will not suffice in anything resembling a stress test, and I may have crazy memory errors flying around right now that I'm not aware of, but this is a benchmark nonetheless. 1.28125V to the memory/graphics controller is apparently the minimum for grandma to check the emails at DDR3-2400, at least with the CPU sample I received (diffused in Germany, made in CHINA?!). I'm going to go ahead and crash Windows now, I'll start with the Calculator and move up to Internet Explorer (that will surely do it). I'll be back!
CPU Voltage - 1.35000
Boost speed - x42 4200MHz
CPU speed - x38 3800MHz
CPU NB voltage - 1.28125 (raised from 1.175 stock)
XMP Memory Profile 1.3
DDR3-2400
10-12-12-31
1.65V
Temperatures in BIOS -
At boot - 24.0 C MB, 27.0 C CPU
After 2 mins - 26.0 C MB, 32.5 C CPU
Well, I typed this up in Notepad; the previous vNB of 1.275 crashed after 5 minutes of Firefoxing. I'm certain that the current vNB will not suffice in anything resembling a stress test, and I may have crazy memory errors flying around right now that I'm not aware of, but this is a benchmark nonetheless. 1.28125V to the memory/graphics controller is apparently the minimum for grandma to check the emails at DDR3-2400, at least with the CPU sample I received (diffused in Germany, made in CHINA?!). I'm going to go ahead and crash Windows now, I'll start with the Calculator and move up to Internet Explorer (that will surely do it). I'll be back!
OP
- Joined
- Mar 27, 2011
- Location
- Soviet Mexico
- Thread Starter
- #40
Not a big deal, good to know that it also has XMP in the bios. I just got a AsRock Extreme 6 board today. However I have had no time to play with it. I need to take pictures of everything first.
Though I did mess with my new ram for about 5 minutes. Seems like that up4 I have dislikes it. Due to not being able to run the higher ram timings. Not a surprise though. for some reason this specific up4 is more uptight about ram settings than the other one was even. Atleast the other board did 2800. It died but it did 2800. Which any A85x board really should be able to support aslong as the chip has a strong enough IMC.
I'll probably give this asrock board a run tomorrow. It should support the 2666 out of the box.
Though I did mess with my new ram for about 5 minutes. Seems like that up4 I have dislikes it. Due to not being able to run the higher ram timings. Not a surprise though. for some reason this specific up4 is more uptight about ram settings than the other one was even. Atleast the other board did 2800. It died but it did 2800. Which any A85x board really should be able to support aslong as the chip has a strong enough IMC.
I'll probably give this asrock board a run tomorrow. It should support the 2666 out of the box.
Similar threads
