Venice OCing thread!

[Duvie]

Ppl I am seeing testing running 1:1 at 260fsbs with cas 2.5-3-3....doesn't seem like isolation to me....drop those dividers to 166-133 ad run at cas 2.5-4-4-10 and then keep trying....I think you are getting hung up by the memory controller..If you gain hundreds more of mhz in cpu speed it will trump those memory settings anyways...

 

[Painman]

My highest PI search was done with mem @ 3-4-4-10 on 100 MHz divider... I retested that 2475 result with the mem brought back up to 2.5-3-3-7 to see if it would break, but it held steady...

Currently Priming the chip at my Winnie's sweet spot (285 * 9) with the 1.57 voltage of course, not the Winnie's 1.53... RAM at 256 2.5-3-3-7... it's early yet but holding steady...

 

[Duvie]

Look at Koy***** (however you spell it) pic...I think those memory settings are tight for even most PC4400-PC4800 modules....He has a nice 2.61ghz but maybe the memory is too high and holding him at that wall....

Maybe not who knows but without isolation many of us will not get the answers we are looking for....We care less about your systems cause chances are we have much different components...we want to know the cpus true limit....What memory setting you get to work later is your deal,...Sorry for being selfish, just being honest...

 

[brentpresley]

Gotta toss my hat into the ring here.

-=General Info=-
CPU Model Number: ADA3500DAA4BP
CPU Code: CBBLE 0512DPAW
CPU Serial Number: forgot to write it down :-(

-=System Info=-
Motherboard: DFI NF4 Ultra-D rev-AA02
Memory: Corsiar XMS 3200C2-PT @ 215.8mhz 2.0 3 3 7
PS: Seasonic Super Tornado 400W

-=Stable OC Info=-
Max Stable OC: 2805mhz
Prime95 LFFTs: haven't run yet, still pushing up the speed
Prime95 SFFTs: XX:XX Time
CPU Temp Under Load: 43C

-=Default Vcore=-
Max OC w/ 1M Super PI results: 2805mhz - 31 Seconds

I've really pushed the Vcore up on this chip (1.60V). It hit a bit of a glass ceiling at about 2.4-2.5GHz that I couldn't get through until i pushed the Vcore up significantly.

Currently I'm using the Super PI 32M test to give me an idea of stability. I'll run Prime95 after I get a good idea what the ceiling is on this chip.

Screen capture can be found here.

 

[Painman]

Originally posted by: Duvie
Look at Koy***** (however you spell it) pic...I think those memory settings are tight for even most PC4400-PC4800 modules....He has a nice 2.61ghz but maybe the memory is too high and holding him at that wall....

Maybe not who knows but without isolation many of us will not get the answers we are looking for....We care less about your systems cause chances are we have much different components...we want to know the cpus true limit....What memory setting you get to work later is your deal,...Sorry for being selfish, just being honest...

I don't think you're being selfish... I'd like this to be an analysis and not a show and tell also, but I also have to remind myself that this testing is more than just for my own info, case in point, I gave up on using 11X divider and went to 10X to beat up on the mem controller some... in my case I was running the RAM at 256 2.5-3-3-7 which already means it was sleepwalking. This is something I know about my RAM, but not necessarily everyone else...

 

[bjc112]

Are all "CBBLE" Venice cores?

 

[housecat]

These results are all good IMO. I'm shooting for 2.5 on stock voltage, something my 3200+ week 51 winny wouldnt do reliably without 1.65-1.7vcore.

I'll have my results up here Monday.

I'm just going to edit it into this post, to not add more trash to sort through in this thread.. there already is alot of banter and not enough results.

 

[Koyanisquatsi]

Originally posted by: Duvie
Look at Koy***** (however you spell it) pic...I think those memory settings are tight for even most PC4400-PC4800 modules....He has a nice 2.61ghz but maybe the memory is too high and holding him at that wall....

Maybe not who knows but without isolation many of us will not get the answers we are looking for....We care less about your systems cause chances are we have much different components...we want to know the cpus true limit....What memory setting you get to work later is your deal,...Sorry for being selfish, just being honest...

I've tried it with loose timings and mem dividers. Memory is not the issue in my case. I really think it's my cooling holding me back. As far as prime95 stability I hit a wall at 2.6ghz with 1.47 volts. To run 2.7ghz prime stable it's taking me 1.65 volts and heat is the limiting factor for me there.

 

[Duvie]

That is not hard to believe...I can run 2.6ghz at 1.47-1.48 now and to get 2.7ghz to run stable it needs 1.62v.....

What kind of temps are you seeing by the way...swings from idle to load would be nice....Havbe you tried running this outside the case in the open air???

 

[user1234]

these are not great initial results, but still better then winchester, which I think on average do no more than 2.5, so maybe venice can do 100-200 or so better. not too bad really

 

[Duvie]

The winchesters did good from week 36-48 then the fvcking floor fell out on them and they are basically dogshite most of them...Look at my old poll....2.6 and above almost accouted for 1/3rd with actually like 29%..... 2.5 range another was 30% on its own so a majority back then got 2.5ghz and higher...Now iti s like a rare 25% even make it to 2.5ghz with the CBBID chips.....

I dont what AMD did to cripple them but they did it...

 

[ozziegn]

man, so far these 3500+ Venice O/Cing results are really disappointing to me. I just ordered one because one of the guys at the OCforums is getting a rock solid, 3.2Ghz from his 3500+ Venice (on 1.60 volts) and I was hoping to get somewhat near the same results. now grant it, he does have a Prometia system which I'm sure helps out alot but I would think that getting 3Ghz+ on high quality air cooling wouldnt be too much to ask for out of these chips.

jeez.... I'm going to be so bumbed if I get these pitiful 2.5 to 2.8Ghz results like what I'm seeing so far in this thread. just pitiful...

 

[brentpresley]

Originally posted by: ozziegn
man, so far these 3500+ Venice O/Cing results are really disappointing to me. I just ordered one because one of the guys at the OCforums is getting a rock solid, 3.2Ghz from his 3500+ Venice (on 1.60 volts) and I was hoping to get somewhat near the same results. now grant it, he does have a Prometia system which I'm sure helps out alot but I would think that getting 3Ghz+ on high quality air cooling wouldnt be too much to ask for out of these chips.

jeez.... I'm going to be so bumbed if I get these pitiful 2.5 to 2.8Ghz results like what I'm seeing so far in this thread. just pitiful...

AMD is speed binning these chips. I guarantee it. If you want 3.0GHz on air, you will have to buy a 4000+ San Diego.

Expecting over 2.8GHz out of Venice is an unreal expectation. Granted, all the tech sites that reviewed these got that good or better, but I would bet they got cherry-picked CPUs from AMD.

 

[ozziegn]

Originally posted by: brentpresley
AMD is speed binning these chips. I guarantee it. If you want 3.0GHz on air, you will have to buy a 4000+ San Diego.

Expecting over 2.8GHz out of Venice is an unreal expectation. Granted, all the tech sites that reviewed these got that good or better, but I would bet they got cherry-picked CPUs from AMD.

exact what is speed binning?

 

[brentpresley]

SPEED BINNING:
All cpus are tested at the factory after they are cut from the silicon. This usually happens before they are mounted onto the pin grid array packages. During this testing, AMD/Intel/whomever can run enough tests to get a good idea what the max stable speed at a certain voltage is for each chip. Due to impurities in the silicon, manufacturing defects, etc., some CPUs are better than others and will run at higher speeds. Naturally, AMD would set these chips aside to sell as their fastest chips (currently FX-55s and 4000+s). If a chip fails at the highest speed, it is tested one step down . . . and so forth.

Speed binning usually occurs when processors are just released and the factory is having trouble meeting demand on it's highest level chips. Later on, when the manufacturing process has been further tweaked, lots of chips will pass at the highest grade. Many more than AMD can sell. Some of these "good" chips will then be sold at lower speed grades, even though they can run at the full 4000+ speed (or sometimes higher). As overclockers, this is what we are all hoping for.

Since the Venice revision of the A64 core contains significant improvements designed specifically to increase yields at higher speeds, we have all been hoping that even the lowest graded Venice chips would overclock well.

 

[TrueWisdom]

Hey Duvie, I sent you a quick PM. Would you mind hitting me back?

 

[user1234]

Originally posted by: brentpresley
SPEED BINNING:
All cpus are tested at the factory after they are cut from the silicon. This usually happens before they are mounted onto the pin grid array packages. During this testing, AMD/Intel/whomever can run enough tests to get a good idea what the max stable speed at a certain voltage is for each chip. Due to impurities in the silicon, manufacturing defects, etc., some CPUs are better than others and will run at higher speeds. Naturally, AMD would set these chips aside to sell as their fastest chips (currently FX-55s and 4000+s). If a chip fails at the highest speed, it is tested one step down . . . and so forth.

Speed binning usually occurs when processors are just released and the factory is having trouble meeting demand on it's highest level chips. Later on, when the manufacturing process has been further tweaked, lots of chips will pass at the highest grade. Many more than AMD can sell. Some of these "good" chips will then be sold at lower speed grades, even though they can run at the full 4000+ speed (or sometimes higher). As overclockers, this is what we are all hoping for.

Since the Venice revision of the A64 core contains significant improvements designed specifically to increase yields at higher speeds, we have all been hoping that even the lowest graded Venice chips would overclock well.

that sounds believable, except that it doesn't explain why the early winchesters were good overclockers like Duvie (a.k.a "the man") said, but later ones are "dog shite" (which I agree with this assertsion since I got a CBBID).

 

[brentpresley]

Can't say why the early winnies were better OCers, but speed binning has been going on in the CPU biz since the 8088 era (as has overclocking). It's possible that AMD was re-tooling the fab or saving the really nice A64 cores for the notebook processors (where a high-speed, lower voltage cpu is critical).

 

[Zebo]

Definity AMD did something with winnys. Week 48 SPAWs hit 2.8 on default Vcore as the norm!!!! And all of a sudden, two weeks later thier manufacturing process gets 400Mhz worse! Not likly. Does'nt work that way, you refine all the time for higher yield of CPUs which intrinsically yeild higher clocks. AMD did "something". Week 50 on were horrible chips, averaging out around 2400 Mhz.

I don't think this is happening here though. Instead AMD speed binning, prolly lower yeild ATM, and you might get real lucky or one that just made the grade. Once a couple month pass then they will have "perfected" process and we might get into a "SPAW" situation...then they cripple them again.

I'm waiting ATM (although frys was out tonight of BI's)

 

[Dadofamunky]

Hehe don't get me wrong I'm not coming from a Tbird @ 1.4. I was just relating what my last AMD based system was. What I'm actually upgrading from now is a P4 2.6C HT @ 3.25ghz with i875 chipset. The only reason for the move is for the pure gaming speed that the AMD64 seems to excell at. Talk of the enhancements in the new Venice core made me cave to the pressure. Fortron Blue Storm 500 24-pin and MSI K8N Neo4 Platinum are ready and waiting for this Venice CPU to arrive.

That's cool. OCing Intels is a world in itself. I've used them for years (Ocd a P3 -700 to 850 Mhz and ran it for almost 5 years as my main machine). You definitely went for the right mobo. If I'd been able to waot a few more weeks I would have gotten PCI-E as well. Oh well, at least I have Socket 939 LOL

 

[bjc112]

Originally posted by: brentpresley
SPEED BINNING:
All cpus are tested at the factory after they are cut from the silicon. This usually happens before they are mounted onto the pin grid array packages. During this testing, AMD/Intel/whomever can run enough tests to get a good idea what the max stable speed at a certain voltage is for each chip. Due to impurities in the silicon, manufacturing defects, etc., some CPUs are better than others and will run at higher speeds. Naturally, AMD would set these chips aside to sell as their fastest chips (currently FX-55s and 4000+s). If a chip fails at the highest speed, it is tested one step down . . . and so forth.

Speed binning usually occurs when processors are just released and the factory is having trouble meeting demand on it's highest level chips. Later on, when the manufacturing process has been further tweaked, lots of chips will pass at the highest grade. Many more than AMD can sell. Some of these "good" chips will then be sold at lower speed grades, even though they can run at the full 4000+ speed (or sometimes higher). As overclockers, this is what we are all hoping for.

Since the Venice revision of the A64 core contains significant improvements designed specifically to increase yields at higher speeds, we have all been hoping that even the lowest graded Venice chips would overclock well.

Good post..

I think it pretty much comes down to this, AMD must realize the if a majority of the chips are hitting 2.4l2.5l2.6 why sell them @ 1.8ghz?

Especially when most of those 2.4's can be had on default volts.

Might as well make some extra money, why would they give away such LARGE amounts of performance? For free

 

[Shimmishim]

Originally posted by: bjc112

Originally posted by: brentpresley
SPEED BINNING:
All cpus are tested at the factory after they are cut from the silicon. This usually happens before they are mounted onto the pin grid array packages. During this testing, AMD/Intel/whomever can run enough tests to get a good idea what the max stable speed at a certain voltage is for each chip. Due to impurities in the silicon, manufacturing defects, etc., some CPUs are better than others and will run at higher speeds. Naturally, AMD would set these chips aside to sell as their fastest chips (currently FX-55s and 4000+s). If a chip fails at the highest speed, it is tested one step down . . . and so forth.

Speed binning usually occurs when processors are just released and the factory is having trouble meeting demand on it's highest level chips. Later on, when the manufacturing process has been further tweaked, lots of chips will pass at the highest grade. Many more than AMD can sell. Some of these "good" chips will then be sold at lower speed grades, even though they can run at the full 4000+ speed (or sometimes higher). As overclockers, this is what we are all hoping for.

Since the Venice revision of the A64 core contains significant improvements designed specifically to increase yields at higher speeds, we have all been hoping that even the lowest graded Venice chips would overclock well.

Good post..

I think it pretty much comes down to this, AMD must realize the if a majority of the chips are hitting 2.4l2.5l2.6 why sell them @ 1.8ghz?

Especially when most of those 2.4's can be had on default volts.

Might as well make some extra money, why would they give away such LARGE amounts of performance? For free

i might be totally wrong...

but i think they did it for the budget conscious consumer...

for example... they produce 99 chips out of 100 capable of 2.8 ghz.... but try to sell them for $1000.... are they going to sell many? some i'm sure...

but lower the speed to 1.8 ghz, top lock the mult at 9x and sell these babies for $130... would people buy these chips? yes!

i'm guessing from the way things work out that they end up selling more lower speed processors (3000/3200+) than high end (4000+/FX55) and end up making more money off the lower speed ones even though they can be clocked to 2.8

then again you have to remember that not everyone overclocks!

 

[dug777]

Originally posted by: NINaudio
Well, it's hard to really find what my results will be as this isn't an official bios release. I would like to wait for that to come out to make a final decision.

Let me ask you guys a question. I'm trying to run LFFT's now and I keep getting an "Illegal Sumout Error", ever seen that before?

more info on this would be great

i get this when i go over 2.2ghz with my 2000+ t-bred, but no other stability problems noticeable

good luck everyone with those venice cores

what do people get out of the fx-55s on air these days? (they are clawhammers right?, and sledgehammer was the s940 fx-51? )

Cheers

 

[housecat]

Originally posted by: Shimmishim

Originally posted by: bjc112

Originally posted by: brentpresley
SPEED BINNING:
All cpus are tested at the factory after they are cut from the silicon. This usually happens before they are mounted onto the pin grid array packages. During this testing, AMD/Intel/whomever can run enough tests to get a good idea what the max stable speed at a certain voltage is for each chip. Due to impurities in the silicon, manufacturing defects, etc., some CPUs are better than others and will run at higher speeds. Naturally, AMD would set these chips aside to sell as their fastest chips (currently FX-55s and 4000+s). If a chip fails at the highest speed, it is tested one step down . . . and so forth.

Speed binning usually occurs when processors are just released and the factory is having trouble meeting demand on it's highest level chips. Later on, when the manufacturing process has been further tweaked, lots of chips will pass at the highest grade. Many more than AMD can sell. Some of these "good" chips will then be sold at lower speed grades, even though they can run at the full 4000+ speed (or sometimes higher). As overclockers, this is what we are all hoping for.

Since the Venice revision of the A64 core contains significant improvements designed specifically to increase yields at higher speeds, we have all been hoping that even the lowest graded Venice chips would overclock well.

Good post..

I think it pretty much comes down to this, AMD must realize the if a majority of the chips are hitting 2.4l2.5l2.6 why sell them @ 1.8ghz?

Especially when most of those 2.4's can be had on default volts.

Might as well make some extra money, why would they give away such LARGE amounts of performance? For free

i might be totally wrong...

but i think they did it for the budget conscious consumer...

for example... they produce 99 chips out of 100 capable of 2.8 ghz.... but try to sell them for $1000.... are they going to sell many? some i'm sure...

but lower the speed to 1.8 ghz, top lock the mult at 9x and sell these babies for $130... would people buy these chips? yes!

i'm guessing from the way things work out that they end up selling more lower speed processors (3000/3200+) than high end (4000+/FX55) and end up making more money off the lower speed ones even though they can be clocked to 2.8

then again you have to remember that not everyone overclocks!

I'm not overclocking my old winchester 3200 once i get it into my 2nd machine..

the venice is going to enjoy a nice, easy 2.5ghz OC though.

 

[NINaudio]

Originally posted by: dug777

Originally posted by: NINaudio
Well, it's hard to really find what my results will be as this isn't an official bios release. I would like to wait for that to come out to make a final decision.

Let me ask you guys a question. I'm trying to run LFFT's now and I keep getting an "Illegal Sumout Error", ever seen that before?

more info on this would be great

i get this when i go over 2.2ghz with my 2000+ t-bred, but no other stability problems noticeable

good luck everyone with those venice cores

what do people get out of the fx-55s on air these days? (they are clawhammers right?, and sledgehammer was the s940 fx-51? )

Cheers

Turns out that at least in my case it was an issue of not having enough vcore. I bumped it up to 1.61v and so far so good at 2750, been running Prime for about 6 hours so far, gonna let it run all day and see what it does. I will be updated my OP in a minute as well.