E8500 @ 4.5Ghz

[Tempered81]

Originally posted by: Rick James
How's 4.75Ghz sound guys?

Holy smokes! Nice work, see how much vcore it takes to pass an hour of Prime or Orthos.
Mine does 6 hours prime at 4.3Ghz using 1.39 vcore.

Here is a goal for you - 5.1Ghz at 1.47v 10 hours prime stable:
http://i36.tinypic.com/2mx0l74.jpg

 

[error8]

Originally posted by: Duvie

Originally posted by: error8

Originally posted by: TidusZ
Skip linpack, its just a heat test not a stability test. I recommend you leave p95 on overnight though if you want to know if its stable or not.

Where did you figured that out? I ran Prime 95 5 hours straight and it gave me no error. Used Linpack for 10 minutes and it failed. Linpack is much more stressful then Prime 95 and because it also gives you higher then normal temps,what does that tell you? That is weaker then Prime?

Exactly.....so prime95 already being an unrealistic real world example stressing a cpu/cache etc in ways that no real world apps uses just points to the fact of why Linpack is not needed....

Prime95 already loads my chips 7-8c higher then anything I have been able to duplicate on it with apps I use....so WHAT DAMN GOOD does linpack heating it another 8-10c have to do for me? except possibly damage it!!!

As long as I have been 24-48 hours prime95 stable with all cores running I have never ran into a stability issue (except with crappy gskill memory when it fails because it was just over-rated shite anyways)

I dont mind OCCT, but bottomline prime95 does just fine.....And take that from a person who does DCC and doesn't get errors, EEU's, etc.

Well if you all have time to test the cpu 2-3 days without using it during this time, then it's just fine by me. I don't have 3 boxes, so whenever I oc one, I can take my work on the other two, so having to keep my computer occupied with prime for so many hours is not a solution. That is why I use Linpack, because it shows me if it's unstable faster, even though I'll never bump in a real world situation where my cpu will shoot up to 86 C like Linpack does.

 

[BTRY B 529th FA BN]

Test your subsystem but using large FFT's then the chip itself by using Small FFT's

I agree with you on the waiting, it sucks. SO while that's also a topic, what does everyone think about using the custom setting and running each cycle for 5-10 min this way it goes through alot of different "K" sizes. ...

 

[BonzaiDuck]

Originally posted by: Rick James
Well there will be no 5Ghz today.

Memory won't take anymore. Even loosened the timing to 6-6-6-18 and it was still a no go. Ohh well. I can live with 4.5Ghz for a few days.

Yo, Rick.

I may or may not have something to offer here, but the flow of info doth go both ways . . . ya know . . .

I'm running an E8600 on an eVGA [nVidia] 780i board. I'd originally OC'd it to 4.1 Ghz, got it Linpack/IntelBurnTest stable (20 iterations) and dropped it back at the same voltages to 4.0. I was using G.SKILL DDR2-1000 2x2GB @ 2.025V underclocked to 800 with tighter latency settings.

Then in January, I discovered these G.SKILL DDR2-900's ("Black-Pi") rated at 4,4,4,12, with some believable customer-reviews showing that you could actually get them to run beyond 900 Mhz at 4,4,4,12 and 1.975V. Still skeptical about the reliability of those results, I grabbed the 4GB kit for about $54. I think they can still be had at the Egg for that price.

Keep in mind I'm on air-cooling -- not a hybrid with TEC like your CoolIt -- and I'm cautious. These Wolfdale E0's sometimes come with defective tJunction sensors -- the skinny (and my personal experience) is that they are stuck at some temperature value, and Intel says they are "not made to report idle temperature values." You didn't say much about your temps, but I'd think from the CoolIt reviews, there's nothing much to worry about. But it explains my caution.

The biggest risk of damage comes from over-volting these suckers. The CPU-VTT (or CPU-FSB) voltage should not be allowed to exceed 1.4V. The Intel "safety-range" is between 0.85 and 1.36V, as I understand. The relevant measurement for comparison purposes is the monitored reading under load shown with CPU-Z, and I'm currently showing 1.272 to 1.28V with a "set" value of 1.33125V.

Right now I'm stress-testing at 4.25 Ghz with the new "Black-Pi" modules. You may want to give those a whirl -- pretty cheap, in my opinion. . . .

 

[Rick James]

Originally posted by: BonzaiDuck

Originally posted by: Rick James
Well there will be no 5Ghz today.

Memory won't take anymore. Even loosened the timing to 6-6-6-18 and it was still a no go. Ohh well. I can live with 4.5Ghz for a few days.

Yo, Rick.

I may or may not have something to offer here, but the flow of info doth go both ways . . . ya know . . .

I'm running an E8600 on an eVGA [nVidia] 780i board. I'd originally OC'd it to 4.1 Ghz, got it Linpack/IntelBurnTest stable (20 iterations) and dropped it back at the same voltages to 4.0. I was using G.SKILL DDR2-1000 2x2GB @ 2.025V underclocked to 800 with tighter latency settings.

Then in January, I discovered these G.SKILL DDR2-900's ("Black-Pi") rated at 4,4,4,12, with some believable customer-reviews showing that you could actually get them to run beyond 900 Mhz at 4,4,4,12 and 1.975V. Still skeptical about the reliability of those results, I grabbed the 4GB kit for about $54. I think they can still be had at the Egg for that price.

Keep in mind I'm on air-cooling -- not a hybrid with TEC like your CoolIt -- and I'm cautious. These Wolfdale E0's sometimes come with defective tJunction sensors -- the skinny (and my personal experience) is that they are stuck at some temperature value, and Intel says they are "not made to report idle temperature values." You didn't say much about your temps, but I'd think from the CoolIt reviews, there's nothing much to worry about. But it explains my caution.

The biggest risk of damage comes from over-volting these suckers. The CPU-VTT (or CPU-FSB) voltage should not be allowed to exceed 1.4V. The Intel "safety-range" is between 0.85 and 1.36V, as I understand. The relevant measurement for comparison purposes is the monitored reading under load shown with CPU-Z, and I'm currently showing 1.272 to 1.28V with a "set" value of 1.33125V.

Right now I'm stress-testing at 4.25 Ghz with the new "Black-Pi" modules. You may want to give those a whirl -- pretty cheap, in my opinion. . . .

Bon,

I have my eyes set on a pair of patriot "Viper" series since i've had such success with the standard DDR2 patriot's.

Temps at idle at 4.75ghz we're looking at 25-27C and 39-41C under full load. I'm not worried about frying it because fry's gives me a 15 day return policy no matter what

 

[BonzaiDuck]

Originally posted by: Rick James

Bon,

I have my eyes set on a pair of patriot "Viper" series since i've had such success with the standard DDR2 patriot's.

Temps at idle at 4.75ghz we're looking at 25-27C and 39-41C under full load. I'm not worried about frying it because fry's gives me a 15 day return policy no matter what

Like I said -- You've got that CoolIt job.

I'm assuming that your "rig" specs showing 1.344V have the voltage reading from CPU-Z?

My "setting" is 1.35V, and it's showing 1.29-something under CPU_Z. If I can get it stable at 4.25 Ghz, I'll drop it back to 4.2 for regular use.

 

[richierich1212]

Originally posted by: Rick James
God this processor rocks. All i had to do is read up alittle on how to configure intel chipsets and BAM. Can't wait to get some faster ram.

Can't believe i wasted money on an AMD setup before this. Psh.

4.2Ghz

True but you are comparing a quad-core processor with a dual-core processor. AMD is still behind Intel, but slowly trying to catch up. (AMD does not have anything dual-core wise to compete with the Wolfdale processors). There is no comparison here, don't kick a guy while he's still down.

Impressive overclock though!

 

[BonzaiDuck]

I keep noticing with these C2D's and C2Q's -- probably what generally applies to all processors: You eventually get to a point where the required voltage increases move upward toward the steep side of an exponential curve. For simple air-cooling, this is as much a problem because of the heat, but for more effective hybrid or water solutions, the problem would be over-volting.

For this E8600 rig, I think my target will be 4.25 Ghz to run at 4.20 at the same settings.

 

[Dadofamunky]

Originally posted by: BonzaiDuck

The biggest risk of damage comes from over-volting these suckers. The CPU-VTT (or CPU-FSB) voltage should not be allowed to exceed 1.4V. The Intel "safety-range" is between 0.85 and 1.36V, as I understand. The relevant measurement for comparison purposes is the monitored reading under load shown with CPU-Z, and I'm currently showing 1.272 to 1.28V with a "set" value of 1.33125V.

Right now I'm stress-testing at 4.25 Ghz with the new "Black-Pi" modules. You may want to give those a whirl -- pretty cheap, in my opinion. . . .

He looks like he's in no danger of overvolting. I've swallowed hard and set mine up at 1.3875V and now I'm 10 Hr. Prime 95 Blend stable at 4.2 GHz. The monitored voltage under load shows as 1.288-1.292V! I believe Idle appears as 1.344V. Then again, I don't think my board is the best. I am getting "decent" results with it though. I'm going down to play with it some more.

 

[OverVolt]

3636Mhz @ 1.344v for me. I'm jealous. Of course I have a $120 E7200 =p

 

[BonzaiDuck]

Originally posted by: Dadofamunky

Originally posted by: BonzaiDuck

The biggest risk of damage comes from over-volting these suckers. The CPU-VTT (or CPU-FSB) voltage should not be allowed to exceed 1.4V. The Intel "safety-range" is between 0.85 and 1.36V, as I understand. The relevant measurement for comparison purposes is the monitored reading under load shown with CPU-Z, and I'm currently showing 1.272 to 1.28V with a "set" value of 1.33125V.

Right now I'm stress-testing at 4.25 Ghz with the new "Black-Pi" modules. You may want to give those a whirl -- pretty cheap, in my opinion. . . .

He looks like he's in no danger of overvolting. I've swallowed hard and set mine up at 1.3875V and now I'm 10 Hr. Prime 95 Blend stable at 4.2 GHz. The monitored voltage under load shows as 1.288-1.292V! I believe Idle appears as 1.344V. Then again, I don't think my board is the best. I am getting "decent" results with it though. I'm going down to play with it some more.

We've crossed paths before, Dado. No matter how much I feel more experienced OC'ing, I find myself feeling sorta stupid sometimes. This has happened on more than one motherboard I've had, but at some point -- going up the VCORE "settings" ladder, I find the reported values lagging farther and farther behind. Maybe I just have a better E8600 -- dunno. Or -- it could be your board, as you say. See -- I'm currently set at 1.3625V; the reported idle value is about 1.32V; the load shows at 1.304V. This would seem to indicate a difference in your board's vDroop and offset, as opposed to mine.

I was happy with this rig at 4.0 Ghz. At 4.20, I'm 60 Mhz ( in DDR terms) short of the memory's spec or rated speed, but I can tighten the latencies to 4,3,4,9 (and may explore tighter settings) -- while the voltage is still 2.0V for the RAM -- bottom of the range for these G.SKILLs.

Also, because of the "stuck" tJunction sensors, I was squeamish about increasing the thermal power. But if the relevant measure is the reading taken from CPU-Z, then the voltage-squared increases from 1.59 to 1.700 -- going from 1.26V to 1.304V. I'm only guessing here, but if temperature rises proportionally, then this would indicate a 7% increase in load temperature. If (big IF) my sensors are reporting correctly that it (tJunction) tops out at 68C with Linpack/IntelBurnTest at a given room-ambient, then I'd expect it to rise to around 72C. But the relevant thermal spec for the E8600 is the tCase temperature -- which (theoretically, and based on 65nm C2D's) is lower by some margin than tJunction. And the tCase thermal spec is 72C.

Of course, anyone here would tell me -- or you -- that trusting these sensor readings (voltage or temperature) is like trying to use cockpit instruments in Fright Stimulator to land your DC3 in a snowstorm and hit the runway where you'd want to if you could see it . . .

EDIT: 9 hours Prime95, "0,0" ; 36 minutes / 10 iterations LinPack "10 of 10" Maximum tCase reported by HW Monitor: 63C @ 70F-room-ambient. Maximum core values: 63C. Load VCORE: 1.288V. To heck with dropping it down to 4.20 Ghz. I'll leave it at 4.25.

 

[M1A]

Rick on your statement below, I would not trust your temp readings if you are loading it with 1.35-1.35V core, Your sensors may not be right.

Bon,

I have my eyes set on a pair of patriot "Viper" series since i've had such success with the standard DDR2 patriot's.

Temps at idle at 4.75ghz we're looking at 25-27C and 39-41C under full load. I'm not worried about frying it because fry's gives me a 15 day return policy no matter what

Just be careful is all I am saying. I did a lot of messing with my e8400 over the past year, given mine is a CO and yours a EO, the load temps seem low..........

Great OC

 

[BonzaiDuck]

Originally posted by: m1a
Rick on your statement below, I would not trust your temp readings if you are loading it with 1.35-1.35V core, Your sensors may not be right.

Bon,

I have my eyes set on a pair of patriot "Viper" series since i've had such success with the standard DDR2 patriot's.

Temps at idle at 4.75ghz we're looking at 25-27C and 39-41C under full load. I'm not worried about frying it because fry's gives me a 15 day return policy no matter what

Just be careful is all I am saying. I did a lot of messing with my e8400 over the past year, given mine is a CO and yours a EO, the load temps seem low..........

Great OC

From my angle, I can't tell. Two things really suck concerning the Wolfdales: the defective sensors on the E0 stepping, and the BIOS revisions that somehow didn't keep up with Penryn revisions to the way the sensors are read. [See an article on the E8500 here at Anandtech around March, '08.]

Then, pertaining to Rick, there's this:

CoolIt Eliminator

If he runs it on "High," with his Penryn C2D OC'd, it may not be so unreasonable. And with everything else, I still couldn't trust a final pronouncement about it from here . . . .

 

[Rick James]

Originally posted by: BonzaiDuck

Originally posted by: m1a
Rick on your statement below, I would not trust your temp readings if you are loading it with 1.35-1.35V core, Your sensors may not be right.

Bon,

I have my eyes set on a pair of patriot "Viper" series since i've had such success with the standard DDR2 patriot's.

Temps at idle at 4.75ghz we're looking at 25-27C and 39-41C under full load. I'm not worried about frying it because fry's gives me a 15 day return policy no matter what

Just be careful is all I am saying. I did a lot of messing with my e8400 over the past year, given mine is a CO and yours a EO, the load temps seem low..........

Great OC

From my angle, I can't tell. Two things really suck concerning the Wolfdales: the defective sensors on the E0 stepping, and the BIOS revisions that somehow didn't keep up with Penryn revisions to the way the sensors are read. [See an article on the E8500 here at Anandtech around March, '08.]

Then, pertaining to Rick, there's this:

CoolIt Eliminator

If he runs it on "High," with his Penryn C2D OC'd, it may not be so unreasonable. And with everything else, I still couldn't trust a final pronouncement about it from here . . . .

Its on high at all times and has proven time after time including cooling a 1.5V X4 Phenom 2 to 33C at idle and 47C at full load. My new mobo should be here tomorrow and will be shooting for 5ghz with my new memory

 

[Dadofamunky]

Originally posted by: BonzaiDuck

We've crossed paths before, Dado. No matter how much I feel more experienced OC'ing, I find myself feeling sorta stupid sometimes. This has happened on more than one motherboard I've had, but at some point -- going up the VCORE "settings" ladder, I find the reported values lagging farther and farther behind. Maybe I just have a better E8600 -- dunno. Or -- it could be your board, as you say. See -- I'm currently set at 1.3625V; the reported idle value is about 1.32V; the load shows at 1.304V. This would seem to indicate a difference in your board's vDroop and offset, as opposed to mine.

I was happy with this rig at 4.0 Ghz. At 4.20, I'm 60 Mhz ( in DDR terms) short of the memory's spec or rated speed, but I can tighten the latencies to 4,3,4,9 (and may explore tighter settings) -- while the voltage is still 2.0V for the RAM -- bottom of the range for these G.SKILLs.

Also, because of the "stuck" tJunction sensors, I was squeamish about increasing the thermal power. But if the relevant measure is the reading taken from CPU-Z, then the voltage-squared increases from 1.59 to 1.700 -- going from 1.26V to 1.304V. I'm only guessing here, but if temperature rises proportionally, then this would indicate a 7% increase in load temperature. If (big IF) my sensors are reporting correctly that it (tJunction) tops out at 68C with Linpack/IntelBurnTest at a given room-ambient, then I'd expect it to rise to around 72C. But the relevant thermal spec for the E8600 is the tCase temperature -- which (theoretically, and based on 65nm C2D's) is lower by some margin than tJunction. And the tCase thermal spec is 72C.

Of course, anyone here would tell me -- or you -- that trusting these sensor readings (voltage or temperature) is like trying to use cockpit instruments in Fright Stimulator to land your DC3 in a snowstorm and hit the runway where you'd want to if you could see it . . .

EDIT: 9 hours Prime95, "0,0" ; 36 minutes / 10 iterations LinPack "10 of 10" Maximum tCase reported by HW Monitor: 63C @ 70F-room-ambient. Maximum core values: 63C. Load VCORE: 1.288V. To heck with dropping it down to 4.20 Ghz. I'll leave it at 4.25.

Yeah, you're in very good shape there. Like Duvie said above, none of these stability tests approximate anything we're likely to do with our systems. Altough he is apparently a G. Skill hater.

I agree completely about the Vdroop differing between boards. I have no doubt of that. I'm astounded at the drop, even with the corresponding BIOS setting disabled. I also dare not jack the voltage up any higher. I also suspect, since my Asus P5E-VM has the low-end Leadtek 883 sound chip, that Asus may have cheaped out on some other aspects of the board, such as the MOSFETS or the northbridge glue logic, it being a uATX part.

At some point I may just say the hell with it and get a small tower and a best-of-breed P45 or P48 board and migrate everything over. Actually it would probably have to be an Asus, since I would really like to retain my XP-64 installation (which I'm actually very happy with) without having to do the whole thing over again. Not for now, though.

That is a pretty awesome OC by Superfreak here, though.

 

[DrMrLordX]

No wonder the E8400, E8500, and E8600s are going for so much. PhII X3s are nice chips and all but when it comes to maximum performance in single-threaded apps, I don't think you can beat these suckers.

 

[exar333]

Originally posted by: poohbear
well the phenom2 720 for the money is a fantastic overclocker. an avg overclock of 3.7ghz w/ three cores @ $145 its a much better deal than an 8400 IMHO. the unlocked multiplier makes overclocking easy & breezy.

Average OC of 3.7? I doubt it. Might be better to say this is a typical overclock with good components.

 

[Rick James]

Originally posted by: ExarKun333

Originally posted by: poohbear
well the phenom2 720 for the money is a fantastic overclocker. an avg overclock of 3.7ghz w/ three cores @ $145 its a much better deal than an 8400 IMHO. the unlocked multiplier makes overclocking easy & breezy.

Average OC of 3.7? I doubt it. Might be better to say this is a typical overclock with good components.

145? Where can you get it for that price? Newegg has it for 169.99. Thats 11 dollars less then a E8500 which BTW beats down the 720.

 

[Flipped Gazelle]

Originally posted by: Rick James

Originally posted by: ExarKun333

Originally posted by: poohbear
well the phenom2 720 for the money is a fantastic overclocker. an avg overclock of 3.7ghz w/ three cores @ $145 its a much better deal than an 8400 IMHO. the unlocked multiplier makes overclocking easy & breezy.

Average OC of 3.7? I doubt it. Might be better to say this is a typical overclock with good components.

145? Where can you get it for that price? Newegg has it for 169.99. Thats 11 dollars less then a E8500 which BTW beats down the 720.

Buy.com has it for $150/shipped.

Check out AT's review... there are many cases where the X3 720 hang w/e8400, and some where X3 tops it pretty handily. Sorry, AT didn't include the e8500 in the comparos.

Basically, depending on the apps used, saying an E8500 "beats down" the X3 720 is as foolish as saying the 720 "beats down" the e8500.

 

[Rick James]

Originally posted by: Flipped Gazelle

Originally posted by: Rick James

Originally posted by: ExarKun333

Originally posted by: poohbear
well the phenom2 720 for the money is a fantastic overclocker. an avg overclock of 3.7ghz w/ three cores @ $145 its a much better deal than an 8400 IMHO. the unlocked multiplier makes overclocking easy & breezy.

Average OC of 3.7? I doubt it. Might be better to say this is a typical overclock with good components.

145? Where can you get it for that price? Newegg has it for 169.99. Thats 11 dollars less then a E8500 which BTW beats down the 720.

Buy.com has it for $150/shipped.

Check out AT's review... there are many cases where the X3 720 hang w/e8400, and some where X3 tops it pretty handily. Sorry, AT didn't include the e8500 in the comparos.

Basically, depending on the apps used, saying an E8500 "beats down" the X3 720 is as foolish as saying the 720 "beats down" the e8500.

I was talking overclocked.

 

[Flipped Gazelle]

Originally posted by: Rick James

Originally posted by: Flipped Gazelle

Originally posted by: Rick James

Originally posted by: ExarKun333

Originally posted by: poohbear
well the phenom2 720 for the money is a fantastic overclocker. an avg overclock of 3.7ghz w/ three cores @ $145 its a much better deal than an 8400 IMHO. the unlocked multiplier makes overclocking easy & breezy.

Average OC of 3.7? I doubt it. Might be better to say this is a typical overclock with good components.

145? Where can you get it for that price? Newegg has it for 169.99. Thats 11 dollars less then a E8500 which BTW beats down the 720.

Buy.com has it for $150/shipped.

Check out AT's review... there are many cases where the X3 720 hang w/e8400, and some where X3 tops it pretty handily. Sorry, AT didn't include the e8500 in the comparos.

Basically, depending on the apps used, saying an E8500 "beats down" the X3 720 is as foolish as saying the 720 "beats down" the e8500.

I was talking overclocked.

Then you should have said so.

I don't know what the average OC's of the X3 720's are at. I'm seeing reports of 3.8-4.0 on the web, but there are so few samples that, IMO, it's a difficult # to trust.

I think that e8500, generally, hits the 4.1-4.4 Ghz zone (air cooliong), so that improves it's standing against the X3 720 by a little.

Again, to say that one "beats down" the other is inaccurate, unless you specify very specific apps.

 

[Tempered81]

x3 720's are pretty good with games compared to e8xxx chips according to that xbit article.

I'd still take the penryn for the oc headroom. deneb just doesn't get there afaik.