• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Love your socket 754?

Hey BigMac, would you mind running Prime95 to at least 5 iterations? It takes me 1.6v at 2.64GHz (220x12) to get a clean Small FFT run.

Just a heads up, I had to drop 60MHz off my overclock, so you might not be as stable as you think :(
 
I'm sure you are correct. I don't push Prime95 too far. As long as I can get it to run stable in my games that's all I'm looking for. I'll run 5 iterations in a couiple of days (when I get some time) and let you know how it goes.
 
BigMacAttack said:
I'm sure you are correct. I don't push Prime95 too far. As long as I can get it to run stable in my games that's all I'm looking for. I'll run 5 iterations in a couiple of days (when I get some time) and let you know how it goes.
Click to expand...

It's a big difference between gaming-stable and Prime95-stable. I can play Oblivion without any crashes at 2.7 but i see a few graphical glitches which is a sign the computer is pushed a bit too far.

Since I also fold for the [H]orde, stability is very important to spit out quality workunits.

 
Xilikon said:
It's a big difference between gaming-stable and Prime95-stable. I can play Oblivion without any crashes at 2.7 but i see a few graphical glitches which is a sign the computer is pushed a bit too far.

Since I also fold for the [H]orde, stability is very important to spit out quality workunits.

Click to expand...

More importantly for the person who is overclocking is the stability of their windows install. Over time an unstable system will begin to f@#k up and eventually you'll need to reinstall.
 
I think this is where I'm going to sit for now. I've got it at 2.64GHz (12x220) with a vCore of 1.6v.

My RAM gave me some surprises at this speed. I put it on a /11 divider, that means 2640/11=240x2=480, so I could get it to DDR480 without raising the FSB any higher.

Now that alone is a nice boost over running it 1:1 (/12 divider) where it would only run at DDR 440, but after I hit my processors wall I went to tweaking my timings. Turns out at vMem at 3.0v I can run at 1T Command Rate :eek:

Screenshot:

Direct link if the image borks

In retrospect it might be better to do 11x240 (2.64GHz) and run my RAM 1:1, so I might give that a try later.
 
I think am going to stay with my good old newcastle because it will do 2.7ghz @ 1.65-1.7V.....Now if I had a lower end s754 2800+3000+ I would def pick up one of these Venice cores and clock it because for 99$ they are great!
 
Xilikon said:
It's a big difference between gaming-stable and Prime95-stable. I can play Oblivion without any crashes at 2.7 but i see a few graphical glitches which is a sign the computer is pushed a bit too far.

Since I also fold for the [H]orde, stability is very important to spit out quality workunits.

Click to expand...
No question about that. If you're folding then stability is paramount. I run normally at sig specs and its as stable as a rock for as long as I want to play. Other functions don't even come close to putting a load on the system like gaming does. I'll run Prime95 on it just to see how it does and how long it can go before crashing.
 
BigMacAttack said:
I'll run Prime95 on it just to see how it does and how long it can go before crashing.
Click to expand...
Cool, if you could also check to see how far you can push it on 1.6v while keeping it Prime stable, I would really appreciate it.

I want to see what yours does at the same voltage as mine. So far it looks like these give out between 2.6 and 2.7GHz...not that I'm complaining, thats still between 200 and 300MHz better than my Newcastle would do (2.4GHz).

I'm also a little surprised that this is the first time I've ever been able to get this ram down to 1T...could it be due to the better memory controller that the Venice cores have?
 
Getting a screen shot of my new Athlon at 2900 MHz was fun but it doesn't run Priime stable anywhere near there.

BIOS Setting:
264 X 10 ~ 2643 MHz
265 X 10 ~ 2659 MHz

I'm Prime stable at 2643 MHz but it starts getting iffy after that. I was Prime stable at 2704 MHz for half an hour with the case and CPU fans at slow speeds. I could probably run here full time if I cranked up the fans but I prefer quiet and stable so I'm going to settle for 2643 MHz.

This E6 Athlon is very similar to my E6 Sempron. The Sempron needed 1.55 volts to be Prime stable at 2560 MHz. This Athlon also needs 1.525 or 1.55 volts to be Prime stable at 2560 MHz. The 512K cache helps when gaming and the 10X multi works better with my board since it tops out at 327 HTT. I was hoping to run stable at 2800+ MHz but I think most users are going to hit the wall at around 2650 to 2700 MHz with air cooling.

athlon26438dg.png
 
What RAM are you running uncleweb? How much voltage did you need to give a 1GB stick to OC with timings that tight?
 
Unknown-One said:
I'm also a little surprised that this is the first time I've ever been able to get this ram down to 1T...could it be due to the better memory controller that the Venice cores have?
Click to expand...
More thank likely.
 
I'm using 1GB of OCZ Platinum.

http://www.ocztechnology.com/products/memory/ocz_el_ddr_pc_3200_platinum

The part number for one stick is supposed to be:
OCZ4001024ELPE

but in CPU-z my stick shows this part number:
OCZ4002048ELDCPE-K

The second part number is usually for 2 sticks in a Dual Channel (DC) kit but for some reason I bought this single stick and ended up with the second part number. It's the same memory either way.

Newegg has them for under $100 / GB.
http://www.newegg.com/Product/Product.asp?Item=N82E16820227209

It's rated at CL2-3-2-5 at DDR400. By increasing the CAS Latency to 2.5 and keeping everything else the same I've found that my memory runs its best at DDR460 - DDR466. I try to adjust the MHz and my memory divider to keep it running as close to here as possible.

A friend bought some PC3200 OCZ Platinum and his does 250 MHz on a Gigabyte board but at 2T. Default voltage for me is about 2.65V and bumping it up above that only causes memory errors. The OCZ website shows the new versions ( 2.x ) can run at 2.8 volts.

I originally had some Crucial Ballistix and I was really impressed by how fast it ran. It was PC-3200 DDR-400 but could run at 290 MHz. I was surprised to find out that Ballistix at 290 MHz and 2T is about the same as OCZ at 233 MHz and 1T. The Ballistix hates 1T on my board and is only good for about 210 MHz, if that. Better to run it as fast as possible at 2T. The Ballistix also liked 256 MHz - 260 MHz at 2.5-2-2-8 2T. The Ballistix liked extra voltage but my board is limited to 2.85 volts.

Update: I just got home and noticed that Prime quit running after 1:57:26. I'm at 2643 MHz and 1.65 volts in a closed case. It's the hottest day of the year here so I might have to drop the voltage to 1.60 and drop the MHz down to about 2600 MHz until it cools down outside. Either that or turn up the fans or head for the cool basement.
 
3000+
3200+
3400+

If you look at the processor details it seems the max temp does not scale from the 3200+ to 3400+ like it does from the 3000+ to the 3200+. Both 3000+ and 3400+ are rated at 65°C while the 3200+ is 69°C. Just thought I'd point that out while I was doing a little research. ;)
 
The max temp rating is just a number. I noticed that the Thermal Power specs scale nicely from 51 Watts to 59 Watts to 67 Watts as the MHz goes up.

There's not a lot of real world difference between any Athlon processors built on the E3 or E6 Venice core. During the manufacturing process chips are speed binned. The ones that can operate at 2400 MHz with the least amount of voltage to remain stable are thrown in the 3400 bin. If they're not quite as good they throw them in the 3200 or 3000 bin. If half the cache is defective then they call them Semprons, etc. etc. Most socket 939 Venice chips also hit the same wall that we're hitting at around 2700 MHz.

AMD would be rich if they could sell every processor they made at the high end price but there is more demand in the mid-range and low end of the market so they end up having to sell Athlon 3400 capable processors as Athlon 3200 or 3000 to meet demand.

It's easy enough for them to change the multiplier from 12 to 11 to 10 or even disable some of the cache to create a Sempron. This allows them to adjust their production to whatever the market wants and is willing to pay without having to come up with a completely different processor design.

Seeing that my Sempron 2600 needs the exact same voltage as my Athlon 3000 to run Prime stable at 2560 MHz says a lot. There's not a hell of a lot of difference in the manufacturing process of these processors within the E3/E6 family.

The differences that you usually hear about are mostly the differences in users motherboard capabilities, memory, power supplies and cooling in their cases. There are also differences in users overclocking abilities, how much time they're willing to put into the project and what they're willing to put up with in terms of noise and stability. For me, if it's not Prime stable it's not stable but you can certainly be game stable at a higher MHz than Prime stable and you can go higher yet if all you're interested in is SuperPi.

When Conroe comes out next month AMD prices are going to drop to compete but I think you can still have a pretty decent computer using a s754 Athlon without having to scrap everything you already own. New motherboard, new DDR2 memory, new PCI-E graphics card, etc., etc. No thanks! :)
 
Indeed. I'll be sticking with my s754 systems awhile longer.

BTW I ran Prime95 last night at 230x12 @1.62v and it went to the 4th stage of the 6th iteration before it locked up. I observed a high temp of 59.5C :eek: during the run. I tried 225x12 @1.5v this AM but it locked during the 2nd stage of the 3rd iteration so I'm going to throw it a little more voltage and run it again tomorrow morning. It only tooped out at 50.5C this AM - much better!
 
Nice writeup, unclewebb and reflecting exactly what i know about the whole cpu manufacturing and limitations. I wanted to add this is the same for every cpu since the pentium 60 days and the dream of every overclocker is to get a cheap cpu which is just marked just because there is a lack of cpu in that speed and this one could be marked as a top end, for some crazy oc results.
 
BigMacAttack said:
I ran Prime95 last night at 230x12 @1.62v and it went to the 4th stage of the 6th iteration before it locked up. I observed a high temp of 59.5C :eek: during the run. I tried 225x12 @1.5v this AM but it locked during the 2nd stage of the 3rd iteration so I'm going to throw it a little more voltage and run it again tomorrow morning. It only tooped out at 50.5C this AM - much better!
Click to expand...
Yea, Prime95 has a habbit of generating some high temperatures.

My chip is acting the same way, 2.7GHz was fine through SuperPi, and was stable enough to benchmark, but it failed Prime almost instantly.

My OC keeps shrinking, lol. But I've finaly locked it down at 2.618 @ 1.6v. It's stable as a rock now, I just let Prime95 run all night and it was still going when I got back to it. I think that may have also set up my Arctic Silver, because my idle temps droped down from 35c to 32c.
 
One of my early overclocking experiments was with a Pentium III - 450 MHz. I was able to boost the FSB from 100 to 133 which gave me a cheap P3 - 600 MHz. Many years later I finally came across a real P3 - 600 and discovered that my chip was actually faster because it was running at 4.5X133 while the real P3 - 600 was running at 6X100. The overclockability of the P3 - 600 was next to nothing. Intel had to boost the core voltage just to get the 600 out the door so a user was lucky to overclock it 5%. That was the end of the line for the original slot 1 P3.

All chips have a theoretical maximum and there isn't a lot of variation from chip to chip within that family until there is a change in the manufacturing process.

The new and improved AM2 Athlon I had recently that ran Prime stable for 24 hours at 2809 MHz is looking better now but with Conroe coming next month, it's just old technology. AMD needs to get the AM2 line down to 65 nm so they can get their clock rates up and remain competitive.
 
Once you find a MHz that your computer likes to run reliably at then you should try and maximize your memory performance at this speed.

My Sempron64 2600 liked running at approximately 2560 MHz because with a memory divider of 11 my memory was able to run near its peak of 233 MHz. If I tried running higher than 2560 MHz, I was forced to use a larger memory divider which killed memory bandwidth and overall performance.

Here's my Athlon 3000 in SuperPi at 2562 MHz:

athlon256215gy.png


When I increase the memory divider from 11 to 12, I have to increase the CPU frequency by almost 50 MHz just to equal my previous performance.

athlon256222se.png


Sometimes you're better off running less MHz. This will allow you to use less voltage and create less heat so you can run fewer fans without having to sacrifice any performance.
 
If anyone wants to read up on the best way to approach an overclock (on the AMD platform), this is it.
 
How high was everyone able to get on the Venice's without upping the volts? I saw Unknown-One got 2.5ghz on his 3400+. The reason I ask is because my board is limited in two ways. First I can only set the reference clock up to 250mhz in the bios and second my voltage increases are limited to +1.7% +3.4% and +5.1% (~+0.071 max). So with the first limitation I was considering getting the 3400+ for the higher multiplier since it's such a cheap upgrade right now. Today I hit 2.4ghz on my 3000+ at stock voltages. I'll have to see if I can hit my max 2.5ghz within the limits of the bios. At this point I'm wondering if I would get any benefit out of the 3400+ at all since it looks like everyone is increasing the voltage a lot more than my board allows to get beyond 2.5ghz.
 
If you really care about the quality of your oc-ing, then you wanna consider a more capable board. I got my DFI LP NF3 250Gb and it was amazing. Shouldn't be that expensive either, some people are selling them in the FS/FT forum.
 
I got to 220x12 @ 1.46v but didn't check to see if it was Prime stable for 24 hrs at that voltage. Your mobo is really restrictive. It might be worth it to get a new one as Emission has said.
 
My Athlon 3000 was Prime stable and error free overnight for 10 hours at 2643 MHz.

I used the maximum BIOS setting of 1.70 volts. CPU-z reported voltage between 1.744 and 1.760 volts at full load.

During a previous test, a BIOS setting of 1.65 volts was not adequate during the heat of the day at 2643 MHz. With some more case and CPU fan noise it might be OK at 1.65 volts.

Room temperature went from 25C ( 77F ) to 22C ( 72F ) overnight druing the test.

CPU temperature was at 55C with the CPU fan at a very quiet 2000 rpm this morning.

The Athlon64 3000, 3200 and 3400 all seem to hit a similar speed so find whatever processor is cheapest and happy overclocking. :)

prime26438gw.png
 
Unknown-One said:
Cool, if you could also check to see how far you can push it on 1.6v while keeping it Prime stable, I would really appreciate it.

I want to see what yours does at the same voltage as mine. So far it looks like these give out between 2.6 and 2.7GHz...not that I'm complaining, thats still between 200 and 300MHz better than my Newcastle would do (2.4GHz).

I'm also a little surprised that this is the first time I've ever been able to get this ram down to 1T...could it be due to the better memory controller that the Venice cores have?
Click to expand...
Here is a pic of my system at 225x12 @1.54v, ram at DDR500 divider, 2.5-3-3-8 2T. Notice how CPU-z reports the ram. Very odd that it does not recognize the DDR500 divider setting. BTW that temp was after the test was shut down. I saw temps of no more than 52.5C during testing.
Primetemp.jpg
 
Welp, looks like my dad just joined the s754 club. His Socket A system had a meltdown about 3 days ago, and since he's out on a business trip I had to put it back together for him.

Now you see, the thing about buying hardware that’s better than your dad's when you're 16, is when I upgrade, your old parts somehow find their way into his computer. I can't eBay most things until two upgrades later >_>

Anyhow, he had me liberate some of my old things from my eBay pile. My old ASRock s754 motherboard, and 3000+ Newcastle just got a new lease on life.

He's just lucky I bought that s754 Venice when I did.

His case also isn’t a tower, so wiring was...interesting to say the least. Pics:



What I can’t figure out is why I can’t keep this cases temperature under control. It has a good airflow pattern now that I’ve added an intake fan to where the PCI slots used to be, yet the second the lid goes on the thing turns into an oven:


Any ideas on that one? The only way to keep it any cooler that I can figure out, is to pop a 120MM hole in the top and add an intake (or exaust?) fan.
 
You might also consider mounting a fan in the side panel. I took my side panel off my old Compaq and put 2 80mm fans in it. Drilled holes for air circulation. Athlon XP 2600 ran hot anyway. A side fan might just be the ticket.
 
Unknown-One said:
What I can’t figure out is why I can’t keep this cases temperature under control. It has a good airflow pattern now that I’ve added an intake fan to where the PCI slots used to be, yet the second the lid goes on the thing turns into an oven:
Click to expand...
The Overture case is also referred to as an "easy bake oven" and is known to be a pile of crap for airflow and having decent temps.

A lot of the reasons why it's a pile of crap is that the old versions didn't have suffice openings to allow air to come in and that the PSU was a low effiency (ie it wasted a lot of energy and made lots of heat) and ran hot. It would then dump that heat right into the case which is a major problem because, in Antec's great wisdom, the PSU is a proprietary one.

In short there isn't a lot you can do cooling wise to improve but one place you can start is by getting a better heatsink on the CPU, the stock cooler is crap especially in that case.
 
Not every motherboard and every BIOS supports the AMD memory dividers correctly.

Here's what CPU-z should be reporting when using an Athlon 3400 with a 12X multi:

DDR200 = CPU / 24
DDR266 = CPU / 18
DDR300 = CPU / 16
DDR333 = CPU / 15
DDR366 = CPU / 14
DDR400 = CPU / 12
DDR466 = CPU / 11
DDR500 = CPU / 10

On your board BigMac, selecting DDR500 seems to be giving you the DDR266 divider of CPU / 18.

2700 MHz / 18 = 150 MHz memory speed or DDR300

If CPU-z is reporting things properly then you're definitely not getting the most out of your memory.

If DDR500 is working properly then at 2700 MHz CPU speed your memory would be working at 2700 MHz / 10 = 270 MHz which Crucial Ballistix would be happy with.

That's why I use A64 Tweaker. It allows you to try different memory dividers on the fly without having to reboot.
 
Well, that could be except that when I select DDR500 and run superpi it shows a 4 second improvement over the DDR333 divider so I suspect that it is running the ram that fast and cpu-z is not reading it correctly. I've run this system on just about every ram divider and it is far faster at DDR466 and DDR500 than it ever was at DDR333 or DDR400.
I'll d/l the latest cpu-z version and see if that clears it up. I ran A64 tweaker awhile back and posted a screenshot - I think I'll go back and see what it said.
Thx for the tip on cpu-z.
 
If SuperPi shows better times with a DDR500 divider compared to a DDR400 divider then it's definitely working. Might want to run another A-B comparison with your new E3 Athlon just to make sure. You can also run the separate latency module that's included with CPU-z. Your latency will definitely drop going from DDR400 to DDR500.

Is there an updated BIOS available for your motherboard? That might help things display correctly in CPU-z if the newest version of CPU-z doesn't fix it. If things are working OK though then it's probably not worth doing a BIOS flash.

BTW, Prime stable at 2700 MHz and 1.568 volts is excellent. I'm going to have to head to the cool basement and crank up my fans to see if my Athlon 3000 can keep up but I don't like my chances!
 
lol - I need to test Prime for 24 hrs. I think it should be good to go, though. I'll also try to see what my max stable oc is at 1.60v

I'm on bios v11 and there is a bios v12 that I could try. I did have v12 installed for a short time so maybe its time to reflash. Its running very well so I really don't feel like messing with it.
I ran 3DMark03 a little while ago and score 12,555 3DMarks @ 1024x768 on the 3200+ Venice with an oc of 248x11, DDR466 divider and X800Pro oc'd to 570 mem/570 core. Now its time to see what this 3400+ Venice will do at DDR500.
 
I don't use the latest F13 BIOS for my Gigabyte board because they screwed up more than they fixed. There are a couple of memory timing settings that get changed after I boot up into Windows when I use F13 but I don't have that trouble with the F12 BIOS so I've stuck with that.

In theory CPUs are speed binned and your 3400 should need less voltage than my 3000 at the same speed.
 
just built a 754 system cause i couldn't pass up the price, i got a question though, i have a 3200+ venice sitting here idling on an nf3 lanparty from dfi, what kind of temps should i be expecting with arctic silver 5 that hasn't burned in yet, i'l idling ~45-46, is that normal for one of these? my athlon xp usually topped out around 48 under load, but that was after months of having arctic silver sitting on it
 
you should be sitting idle in the mid to low 30's

what kind of airflow do you have through your case? Can you give us a number of intake and exaust fans, aswell as their location?
 
You must log in or register to reply here.
Back
Top