Another stubborn X4 945 (125W) C2 Overclocker

[Anonaru]

Good to know! Prime95 is still chugging away almost an hour later. Max temp in HWM was 56ºC during those 800,000k tests but it only peaked there. It hovered around 53-54ºC in those tests after peaking! I may try and go up more... we'll see! Of course my HTlink and CPU-NB frequencies are at 2000 so that may be part of the reason why I'm not having issues. It seems like anytime I go up on those above 2000 I get rounding errors galore in Prime95, even with tweaking voltages.

It would be better to run something like 3.65 or 3.7 core and 2600 on your NB if you could help it. HT doesn't matter, 2000 or match NB doesn't seem to change much on these parts (I typically try to keep it around 2000 on Deneb, just a superstition of mine). You'll definitly notice higher benching scores if you can get a stable OC around your settings WITH a 2.6~ NB. The C2 controllers are weak, so going much higher can be a bet meh. I could get my own C2 up to 2830 ish, but it required more effort than it was worth

 

[Whiskey11]

It would be better to run something like 3.65 or 3.7 core and 2600 on your NB if you could help it. HT doesn't matter, 2000 or match NB doesn't seem to change much on these parts (I typically try to keep it around 2000 on Deneb, just a superstition of mine). You'll definitly notice higher benching scores if you can get a stable OC around your settings WITH a 2.6~ NB. The C2 controllers are weak, so going much higher can be a bet meh. I could get my own C2 up to 2830 ish, but it required more effort than it was worth

I can give it a try again but I'm hesitant to run the CPU clock speed down if raising the CPU-NB frequency isn't going to gain frames in FSX as I know dropping CPU clock speed will drop frames. At 3.675GHz I'm barely getting playable frames in major cities due to the extremely CPU bound nature of the game. I plan on loading it up here in a moment to see if 3.75 GHz is going to be significantly better. Every up in CPU clock speed has increased FPS in that game (yeah it's that CPU bound). Last time I played with the CPU-NB frequency I tried a decent amount of settings and Prime95 would get rounding errors almost immediately after starting the test on at least one core at pretty much any setting over 2000. I don't know if it's just my chip or if I'm that incompetent (likely ) and can't figure it out! Only thing to do is keep on tweaking!

 

[Anonaru]

I can give it a try again but I'm hesitant to run the CPU clock speed down if raising the CPU-NB frequency isn't going to gain frames in FSX as I know dropping CPU clock speed will drop frames. At 3.675GHz I'm barely getting playable frames in major cities due to the extremely CPU bound nature of the game. I plan on loading it up here in a moment to see if 3.75 GHz is going to be significantly better. Every up in CPU clock speed has increased FPS in that game (yeah it's that CPU bound). Last time I played with the CPU-NB frequency I tried a decent amount of settings and Prime95 would get rounding errors almost immediately after starting the test on at least one core at pretty much any setting over 2000. I don't know if it's just my chip or if I'm that incompetent (likely ) and can't figure it out! Only thing to do is keep on tweaking!

If the overall performance of the chip is better when you do both a CPU and NB OC than simply an OC to the CPU frequency, why would you think that just increasing the frequency is better?

I guarentee you that your frame rates at 3.7 / 2000 are lower than 3.65 / 2600, and probably noticably slower than 3.7/ 2600 haha

 

[Whiskey11]

If the overall performance of the chip is better when you do both a CPU and NB OC than simply an OC to the CPU frequency, why would you think that just increasing the frequency is better?

I guarentee you that your frame rates at 3.7 / 2000 are lower than 3.65 / 2600, and probably noticably slower than 3.7/ 2600 haha

Good question and worth investigating... I just wish I could figure out why it is so touchy and doesn't like going higher than the default 2000 aside from being a C2 and being ultra finicky!

 

[Anonaru]

Good question and worth investigating... I just wish I could figure out why it is so touchy and doesn't like going higher than the default 2000 aside from being a C2 and being ultra finicky!

No other reasons available haha. Your NB shouldn't get in the way of your final OC, but you may need to drop your RAM a multiplier in order to get a substantial boost to your NB (Even then, it really should run 2600 / 1,700MHz~ ish ram without any trouble.. Mine could go significantly higher if I was really mean to it )

Get as high as you can multi-only.. Hopefully you found a stable point, and I can try to get your NB up there too

 

[Whiskey11]

No other reasons available haha. Your NB shouldn't get in the way of your final OC, but you may need to drop your RAM a multiplier in order to get a substantial boost to your NB (Even then, it really should run 2600 / 1,700MHz~ ish ram without any trouble.. Mine could go significantly higher if I was really mean to it )

Get as high as you can multi-only.. Hopefully you found a stable point, and I can try to get your NB up there too

Tried the CPU-NB up to 2500 (10x multiplier) and HT-Link at 2500 (10x multiplier) with CPU-NB VID voltage set to 1.275V and after about a minute I got a rounding error in Prime95... dropped the HT-Link back to 2000 and upped the CPU-NB VID voltage to 1.3V (Supposed max recommended?) and Prime95 is chugging away. I'm a little concerned though as the core temp is already up to 53º-54ºC and we haven't reached the 800,000 tests yet. It's only been running 8 minutes though...

 

[Anonaru]

Tried the CPU-NB up to 2500 (10x multiplier) and HT-Link at 2500 (10x multiplier) with CPU-NB VID voltage set to 1.275V and after about a minute I got a rounding error in Prime95... dropped the HT-Link back to 2000 and upped the CPU-NB VID voltage to 1.3V (Supposed max recommended?) and Prime95 is chugging away. I'm a little concerned though as the core temp is already up to 53º-54ºC and we haven't reached the 800,000 tests yet. It's only been running 8 minutes though...

Aye, running it up higher will indeed increase your temps, though not by much. You will probably peak around 57-58 if you were at 56 before, which is still ok (But your thermal ceiling).

If you can hold these settings, you have found your thermal ceiling, and more than likely just about your OC ceiling as well. Definitly not bad for your part

 

[Whiskey11]

Aye, running it up higher will indeed increase your temps, though not by much. You will probably peak around 57-58 if you were at 56 before, which is still ok (But your thermal ceiling).

If you can hold these settings, you have found your thermal ceiling, and more than likely just about your OC ceiling as well. Definitly not bad for your part

Of course my computer likes proving me wrong, almost immediately after I posted this the temps started dropping. I'm still not in the 800,000 tests but it dropped down to 51ºC. It is just now starting the 800,000 so we'll see! Stay tuned! (Temp is actually falling to around 50ºC in the 800,000k tests? that's a head scratcher!)

A part of me is really thinking about adding another radiator to this loop... it's already a 240 or a 120x2 Radiator/Reservoir/Pump combo from Swiftech, but surely another 120 radiator would help keep temps from getting too hot right? Better fans would to but the way it is INSIDE the case the RAM retainer clips are blocking going to a push/pull configuration which is highly annoying. I could probably fit a slim fan in there but they are roughly half the CFM of the fans on the top. I might be able to hack a 120's outer rim off to do it too... Probably a better question for the guys over in the watercooling forums right?

 

[Anonaru]

Of course my computer likes proving me wrong, almost immediately after I posted this the temps started dropping. I'm still not in the 800,000 tests but it dropped down to 51ºC. It is just now starting the 800,000 so we'll see! Stay tuned! (Temp is actually falling to around 50ºC in the 800,000k tests? that's a head scratcher!)

A part of me is really thinking about adding another radiator to this loop... it's already a 240 or a 120x2 Radiator/Reservoir/Pump combo from Swiftech, but surely another 120 radiator would help keep temps from getting too hot right? Better fans would to but the way it is INSIDE the case the RAM retainer clips are blocking going to a push/pull configuration which is highly annoying. I could probably fit a slim fan in there but they are roughly half the CFM of the fans on the top. I might be able to hack a 120's outer rim off to do it too... Probably a better question for the guys over in the watercooling forums right?

Probably a better question for them, yes. If you're on a full custom loop, you should probably be able to get your temps down a few more degrees than you have.. I'm not sure if some fan placement tweaks are in order, or if its going to be more substantial such as adding another block.. You're cooling your GPUs on this loop?

Edit: I wouldn't put in low CFM fans in a push / pull, if they have half the CFM the benefits would be nearly impossible to measure. Replacing your current fans with higher CFM ones, improving case airflow, and making sure your rad isn't caked in dust n crap are good "right now" checks to make. Adding another rad could be nice, or it could bring you something miniscule like 2c if surface area isn't your problem

 

[Whiskey11]

Probably a better question for them, yes. If you're on a full custom loop, you should probably be able to get your temps down a few more degrees than you have.. I'm not sure if some fan placement tweaks are in order, or if its going to be more substantial such as adding another block.. You're cooling your GPUs on this loop?

Edit: I wouldn't put in low CFM fans in a push / pull, if they have half the CFM the benefits would be nearly impossible to measure. Replacing your current fans with higher CFM ones, improving case airflow, and making sure your rad isn't caked in dust n crap are good "right now" checks to make. Adding another rad could be nice, or it could bring you something miniscule like 2c if surface area isn't your problem

It's not quite a "all in one" but it isn't a full custom either. The Swiftech kit includes a Radiator/Reservoir/Pump in one unit but it uses standard barb fittings and hose for customization and expandability. The Appogee HD block is expandable to include other loops as well in parallel rather than in series. Right now it's only a CPU cooler (see first post for photo) and the unit is brand new. Less than a week old.

The case, a Zalman GS1000 isn't known for being a good cooling case but it isn't stock either. Right now I have:
1x140mm Swiftech Helix fan (75 CFM) in the bottom pulling air in to the case
1x120mm Zalman Quiet Fan (Can't find specs, assuming 40-50 CFM) in the top rear of the case
2x120mm Swiftech fans (77 CFM) as intakes for the radiator on the top of the case

I also have one of the Zalman 120mm fans sitting at the ledge between the PSU at the bottom of the case pointed at the GPU to try and get cool air over it.

I did find out that I have my intake fans for the radiator in the wrong fan connectors and are apparently downrated to 5V instead of the full 12V... I'm going to fix that! The rest of the fans are board mounted and should be running at 100% speed.

I'm also debating adding a 140mm fan under the DVD drive in the 3 front drive bays to pull air in. I would also like to cut the fan grills out of the top of the case (planned for tomorrow when it's at least decent outside) so the radiator fans are less obstructed.

I'll get a picture of the inside with it all in there.

EDIT: Ohh yeah, definitely running at full speed on those intake fans now! A little more noise but nothing unbearable! Re-Running Prime95 to see if there is a drop in temp and potentially more room for OCing!!

EDIT2: Looks to be roughly a 7ºC difference between 5V and 12V and the sound, while ever present, isn't THAT bad. It's at least air moving noise and not a high pitched whine. Now the question is... how much more is 7ºC in terms of clock speed?

 

[Anonaru]

It's not quite a "all in one" but it isn't a full custom either. The Swiftech kit includes a Radiator/Reservoir/Pump in one unit but it uses standard barb fittings and hose for customization and expandability. The Appogee HD block is expandable to include other loops as well in parallel rather than in series. Right now it's only a CPU cooler (see first post for photo) and the unit is brand new. Less than a week old.

The case, a Zalman GS1000 isn't known for being a good cooling case but it isn't stock either. Right now I have:
1x140mm Swiftech Helix fan (75 CFM) in the bottom pulling air in to the case
1x120mm Zalman Quiet Fan (Can't find specs, assuming 40-50 CFM) in the top rear of the case
2x120mm Swiftech fans (77 CFM) as intakes for the radiator on the top of the case

I also have one of the Zalman 120mm fans sitting at the ledge between the PSU at the bottom of the case pointed at the GPU to try and get cool air over it.

I did find out that I have my intake fans for the radiator in the wrong fan connectors and are apparently downrated to 5V instead of the full 12V... I'm going to fix that! The rest of the fans are board mounted and should be running at 100% speed.

I'm also debating adding a 140mm fan under the DVD drive in the 3 front drive bays to pull air in. I would also like to cut the fan grills out of the top of the case (planned for tomorrow when it's at least decent outside) so the radiator fans are less obstructed.

I'll get a picture of the inside with it all in there.

EDIT: Ohh yeah, definitely running at full speed on those intake fans now! A little more noise but nothing unbearable! Re-Running Prime95 to see if there is a drop in temp and potentially more room for OCing!!

EDIT2: Looks to be roughly a 7ºC difference between 5V and 12V and the sound, while ever present, isn't THAT bad. It's at least air moving noise and not a high pitched whine. Now the question is... how much more is 7ºC in terms of clock speed?

I was going to say that you'll see a noticable diff. with that change haha.

Good! It means you have a much larger cushin to work with, and water cooling resists temp changes on the top end much better than air coolers do.. However, your C2 chip isn't going to go much higher-- you'll have to find that limit yourself, each chip is different once you break into the 3.7 cloud

 

[Whiskey11]

I was going to say that you'll see a noticable diff. with that change haha.

Good! It means you have a much larger cushin to work with, and water cooling resists temp changes on the top end much better than air coolers do.. However, your C2 chip isn't going to go much higher-- you'll have to find that limit yourself, each chip is different once you break into the 3.7 cloud

My only gripe about going higher is that it is going to require under rating the RAM frequency again. I suppose I could get it back up by using the Multiplier on the CPU to get the proper RAM frequency and still maintain a higher clock speed. Off to go play!

 

[Anonaru]

My only gripe about going higher is that it is going to require under rating the RAM frequency again. I suppose I could get it back up by using the Multiplier on the CPU to get the proper RAM frequency and still maintain a higher clock speed. Off to go play!

There's always a "best" combination between multi and FSB. For my Deneb part, it was some ridicilous number, like 227 (as in weird ridicilous) or something. Some people find this sweet spot much lower (Like 208) and I've seen people find their max OC with a FSB as high as 260 (Not that I reccomend this).

Good luck, happy sailing good sir Don't forget the settings you were holding at before you went to play

 

[Whiskey11]

There's always a "best" combination between multi and FSB. For my Deneb part, it was some ridicilous number, like 227 (as in weird ridicilous) or something. Some people find this sweet spot much lower (Like 208) and I've seen people find their max OC with a FSB as high as 260 (Not that I reccomend this).

Good luck, happy sailing good sir Don't forget the settings you were holding at before you went to play

Funny you should say that... 15X is my max multiplier for the CPU since it's a non-BE CPU. I basically can only tune the OC with FSB. I was typing this as an edit to my post above when you posted:

Tried a couple of other settings and nothing wanted to POST correctly (Getting the MoBo to give me the whole "Boot Failures due to overclocking" message). So I took it back to 250x15 on the FSB, and jumped up 5 more on the FSB to 255 and vcore to 1.5V to see if it would post and Prime95 and it has passed an hours worth of Prime95 Blend with a max temp of 53ºC but averaging around 50ºC through all the testing. My ram is OCed on speed now though to 1359 so I'm not sure if I can go any higher before dropping the DRAM multiplier down to 4x from the 5.33x. I'm wondering if there isn't an issue with going beyond 260 FSB regardless of final clock speed?? I'm relatively new to OCing so I'm not sure. Even at 14.5 on the MP and a 260 FSB wouldn't post even at 1.525V vcore. I'm still running Prime95 in the background here but I can't say I'm disappointed with 3.825GHz and the temps. Higher and higher I go, where I'll stop, nobody knows!

I would like to run this overnight to make sure it is going to be stable for longer than an hour and change but I'm hopeful that it will. When I started this post I felt kind of like getting 3.675GHz to be stable was going to be a huge accomplishment and having it run for 9 hours without error was nice. I hope to have this 3.825GHz meet the same standard. I'm just not sure I can sit here and surf the internet until I crash tonight! I'm just glad I've got the next three days off (interview with a Police Department on Wednesday but otherwise free to tweak away!).

I also have all the previous settings written down on a piece of paper nearby should I need them.

I'm still very curious on what the record for a 125W C2 945 is for overclocking using conventional air/water setups!

 

[Anonaru]

Funny you should say that... 15X is my max multiplier for the CPU since it's a non-BE CPU. I basically can only tune the OC with FSB. I was typing this as an edit to my post above when you posted:

Tried a couple of other settings and nothing wanted to POST correctly (Getting the MoBo to give me the whole "Boot Failures due to overclocking" message). So I took it back to 250x15 on the FSB, and jumped up 5 more on the FSB to 255 and vcore to 1.5V to see if it would post and Prime95 and it has passed an hours worth of Prime95 Blend with a max temp of 53ºC but averaging around 50ºC through all the testing. My ram is OCed on speed now though to 1359 so I'm not sure if I can go any higher before dropping the DRAM multiplier down to 4x from the 5.33x. I'm wondering if there isn't an issue with going beyond 260 FSB regardless of final clock speed?? I'm relatively new to OCing so I'm not sure. Even at 14.5 on the MP and a 260 FSB wouldn't post even at 1.525V vcore. I'm still running Prime95 in the background here but I can't say I'm disappointed with 3.825GHz and the temps. Higher and higher I go, where I'll stop, nobody knows!

I would like to run this overnight to make sure it is going to be stable for longer than an hour and change but I'm hopeful that it will. When I started this post I felt kind of like getting 3.675GHz to be stable was going to be a huge accomplishment and having it run for 9 hours without error was nice. I hope to have this 3.825GHz meet the same standard. I'm just not sure I can sit here and surf the internet until I crash tonight! I'm just glad I've got the next three days off (interview with a Police Department on Wednesday but otherwise free to tweak away!).

I also have all the previous settings written down on a piece of paper nearby should I need them.

I'm still very curious on what the record for a 125W C2 945 is for overclocking using conventional air/water setups!

Really, there's no hard cap for your FSB. A record would be hard to find in this regard.. FSB max is limited by both CPU and motherboard, and sometimes by a pretty big margin.. Motherboard A can get to 240, while motherboard B can get up to 275, for example. You wanna keep your x15 multiplier, and while your RAM is OC'ed from the 1300 setting, if you treat it properly you shouldn't have to further downclock it (If it is rated for 1600 operation). You may have to adjust timings accordingly and perhaps give it a lil extra voltage (Or simply manually set the voltage)

 

[Whiskey11]

Really, there's no hard cap for your FSB. A record would be hard to find in this regard.. FSB max is limited by both CPU and motherboard, and sometimes by a pretty big margin.. Motherboard A can get to 240, while motherboard B can get up to 275, for example. You wanna keep your x15 multiplier, and while your RAM is OC'ed from the 1300 setting, if you treat it properly you shouldn't have to further downclock it (If it is rated for 1600 operation). You may have to adjust timings accordingly and perhaps give it a lil extra voltage (Or simply manually set the voltage)

Well the MoBo says 500 FSB is the max so that's the max right?

No I'm not serious about that!


The RAM is only rated to 1333 so I'm impressed it's managed to live this long. It also seems that it is very intolerant of any other timings then the 8-8-8-21 timings even 9-9-9-24 which is what the MoBo defaults them to when at 1333 speed. They are only rated for 1.5V but the MoBo defaults to a 1.6V rating and wont let me go any lower... a mistake I most likely made when comparing how compatible this memory was with this MoBo. Regardless, the memory has been running at 1.6V for what is rapidly approaching 4 years.

 

[Anonaru]

Well the MoBo says 500 FSB is the max so that's the max right?

No I'm not serious about that!


The RAM is only rated to 1333 so I'm impressed it's managed to live this long. It also seems that it is very intolerant of any other timings then the 8-8-8-21 timings even 9-9-9-24 which is what the MoBo defaults them to when at 1333 speed. They are only rated for 1.5V but the MoBo defaults to a 1.6V rating and wont let me go any lower... a mistake I most likely made when comparing how compatible this memory was with this MoBo. Regardless, the memory has been running at 1.6V for what is rapidly approaching 4 years.

Don't worry about that RAM voltage, though that may be good for ya. Most 1333 ram will run at 1600, so don't be too worried as long as it isn't causing you noticable errors If you get errors, it doesn't hurt to knock it back a multi and try again, but hey.

The theoretical max frequency on my Deneb CPU is something like 14.5GHz, but that doesn't mean it is going that high. Just like your theoretical max is 500MHz FSB, but you're lucky if you get it to post at 260 <.< (Good god would having a 14GHz part be fudging nuts)

 

[Whiskey11]

Don't worry about that RAM voltage, though that may be good for ya. Most 1333 ram will run at 1600, so don't be too worried as long as it isn't causing you noticable errors If you get errors, it doesn't hurt to knock it back a multi and try again, but hey.

The theoretical max frequency on my Deneb CPU is something like 14.5GHz, but that doesn't mean it is going that high. Just like your theoretical max is 500MHz FSB, but you're lucky if you get it to post at 260 <.< (Good god would having a 14GHz part be fudging nuts)

Well I let Prime95 run for a long time and managed to go 6hrs and 17 minutes before coming up with a rounding error. It says it was a rounding error then says it was a hardware failure detected. Six hours is a long dang time to run full tilt. I'm not sure what the issue is or if something needs more voltage or something. Max temp recorded was 54ºC so I don't think it was that for an issue.

Any thoughts?

 

[Anonaru]

Well I let Prime95 run for a long time and managed to go 6hrs and 17 minutes before coming up with a rounding error. It says it was a rounding error then says it was a hardware failure detected. Six hours is a long dang time to run full tilt. I'm not sure what the issue is or if something needs more voltage or something. Max temp recorded was 54ºC so I don't think it was that for an issue.

Any thoughts?

Could be any number of things. What test was the worker running when it failed? With that information it could be possible to determine the most likely cause (I'm thinking IMC or RAM personally)

Normally I say that is pretty darn good, but my Deneb rigs almost always let me down if they didn't make 12 hours in P95 by rewarding my gaming with a BSOD a couple horus in :S

 

[Whiskey11]

Could be any number of things. What test was the worker running when it failed? With that information it could be possible to determine the most likely cause (I'm thinking IMC or RAM personally)

Normally I say that is pretty darn good, but my Deneb rigs almost always let me down if they didn't make 12 hours in P95 by rewarding my gaming with a BSOD a couple horus in :S

The "results.txt" log says the other workers were on the 160k self test. I've long since closed that Prime95 screen. I tried last night to add CPU-NB voltage and going to 1.325V caused 2 to fail immediately (rounding error) so I thought maybe I was over volting, and going to 1.275V caused 2 to fail immediately (rounding error) as well. I've knocked the memory multiplier down so the frequency is 1020 and I'll re-run the tests tonight with the CPU-NB voltage back at 1.3V.

I've also knocked the DRAM voltage down from 1.650V to 1.6V (default). Six hours isn't too bad though, I flew from Laguardia (averaging about 20 frames per second in the cockpit and 25-30 on the exterior) to Orlando International last night (averaged 25-30 in cockpit and 40 on the exterior) and the new clock speed has made it at least playable. I'm going to continue to tweak the game as well as the clock speed. I didn't think I would get this far and even be 6 hour stable in Prime95 so I can't really complain!