Jump to content
Sign in to follow this  
Angry_Games

tight timings? loose timings? dividers? 1T or 2T? 754 or 939? (read he

Recommended Posts

4400+'s have never been great clockers...I've had 4 of them in my hands over time and all of them would do 2600Mhz, a couple would do 2700, one would do 2750, and that's that.

 

I've basically scaled all of my AMD clocks back to 2600Mhz as there's really no reason to go beyond this for gaming, nor even encoding. Saving a couple of seconds or getting a couple of fps more is not worth the hassle of the overclock (unless you have a chip that damn well will overclock beyond that without voltage increase and no failures that take 15 days to pin down just to get that extra 100Mhz).

 

The chip I used in the first post also is one of the best single core cpu's I've ever owned, and all of my 4400+'s were a distant, far cry away from this monster when it came to overclocking.

 

One of the most important rules in overclocking is NEVER judge your own overclocks based on someone else's results. If I took your poor overclocking 4400+, I might be able to make it do 2860Mhz with only 1.45v. I might not even be able to get 2500Mhz out of it with 1.7v.

 

I think you got the idea though now that you know dividers and timings are really a moot point in real world performance (trust me, I've been testing this repeatedly for 2+ years now)

 

Now i need to find exactly where to post for OC'ing help

 

this is the proper section, but make sure you read the OC Guide up and down, inside and out, as much as you possibly can. ALL the answers are in that guide, as well as in the stickies in this section, and then there's the Search function within this section that I guarantee has multiple threads that cover any questions you might have. And of course if all else fails, you can ask in a new thread (but try to follow my advice before you go that route)

Share this post


Link to post
Share on other sites
Guest sutahz

Yeah I've been looking around these forums for OC tips. Also used this guys guide http://icrontic.com/articles/a64_cpu_clock which says a lot of the same things really. He also has a small bar graph showing stepping code avg OC and Max Avg OC (E6 has about 2.7GHz OC Max, as you said as well). So I guess I was HOPEING for more of an OC. I think now I'm comming to accept that 2.58-2.6GHz is good for me (1.64V pushed my Tcase to 55-56 [my cpu is only rated for Tcase 55C] and my PWMIC to 59-60, so both temps maxed out. that was 238x11. Now im sitting at 233x11 1.55V and getting 49-50 Tcase and 55C PWM {much more acceptable, w/o much MHz loss}.)

And yes, i read this whole tread word for word. Im sure this question was answered but I'll ask it anyways. Since memory speed doesnt matter so much (ie 166MHz and up is good), and my cpu's ceiling at 1.55V is ~2.6Ghz, whats most likely going to give me my best results, using a high FSB/HTT (ie 260-270 cause my ram will happily go there) and a low multi, or running a 10 or 11 multi w/ lower FSB/HTT

===

After a lot of reading (and my own laziness) I dont expect 24hr prime95 stability either, 3-4hrs is good enough for me (cause prime95 is one of the most stressful things on your cpu, gameing doesnt push it that hard.. right?). And Core0 always fails first... darn weakest link

Share this post


Link to post
Share on other sites

the better (slightly) of two scenarios (at matching clock speeds) is the lower timings on a lower or no divider.

 

This is always going to be true.

 

However, if you can only get for example 2520Mhz with tight timings and no divider, but you can nail 2640Mhz with tight timings AND a divider, or looser timings and no divider, or even looser timings and a 180/166 divider, your cpu speed is the more important thing here, so 2640Mhz is better than 2520Mhz.

 

the one thing you MUST do though is check for yourself. Run all the apps and games at one setting, then try another setting and check the performance again.

 

I feel like you'll find your results almost exactly like the chart in this thread...hardly any difference, but raw cpu speed is what the difference will come down to ;)

 

also, if you want a killer clocking cpu, look around for one of the Opteron 160, 165, 170, etc. (dual cores)

 

check the 3Ghz thread that is also a sticky and see what all of those guys are using.

Share this post


Link to post
Share on other sites
Guest sutahz

First of all thank you for:

Doing all this testing, and writing up your results

responding to my posts quickly

Now, I very recently read your DDR2 post, which states that DDR2 was needed to get more info to the cpu faster (CPU is fast and smart, but has only short term memory, ram is its long term memory, and of course the cow milking analogy), so the speed of the ram is important. But in this post you basically say the speed of the ram isnt important... hence my bit of confusion. ALSO, as i said im only looking for real world results (and you say 2.6GHz is plenty for a A64), so from your readings, testing, experience, whats a good CPU match for my 8800GTS 640MB running 1680x1040 (and for any games that support dual monitors, i have a 21" 1600x1200 [1928x1378?max] monitor standing by)? Is 2.2GHz enough (STALKER did stutter some for me, hence why i moved on to OC'ing this rig), or 2.4GHz?

Thanks in advance, and after reading these threads, i did look at the price of a opteron 170 and I'd have no qualms aginst swapping out CPUs)

Share this post


Link to post
Share on other sites

don't judge STALKER in your performance testing...it's a hairy balls engine that while looking great, runs like crap even on the hardest of the hardcore rigs.

 

Stick to Half-Life2 engine, Unreal engine, Quake engine, Farcry engine, Battlefield2 engine (though BF2 is still a buggy hunk of crap).

 

I've used both 2x 7950GT SLI + 2600Mhz AM2 X2 + 2GB RAM, and 8800GTS 320MB + 2600Mhz AM2 X2 + 2GB RAM and ALL games play at 1600x1200 w/4xAA/16xAF EXCEPT Unreal3 engine games like Rainbow Six: Vegas.

 

All the Half-Life2 engine games easily did 8xAA, and I've been able to do up to 16xAA on some games.

 

STALKER like all other modern video games (made within the last two years or newer) relies heavily on your video card, not so much on your cpu.

 

Praz did a test with an 1800Mhz Sempron AM2 + X1900XT and a 2400Mhz X2 + X1900XT and in FEAR there was less than 2% difference in game performance.

 

I've done the same tests myself and get the same results.

 

DDR2 is just a faster way to get data in and out of the cpu. It really isn't all that much of a benefit from DDR1, but it will show itself to be better in time just the same as DDR1 did to SDRAM.

Share this post


Link to post
Share on other sites

Hi overclockers,

I had one doubt regarding the memory divider. Since I am unable to put it in words, please check the screen shot of angry's comparison of the one with and without divider pic.

 

divider_doubt.jpg

 

This is with respect to the Centon RAM in particular.

 

If the DRAM frequency Set is 200 and the divider is 01/01, then how can the FSB Bus Frequency be 275 MHz.

 

If anybody would clarify this to me, may be then I would understand the divider concept. Me feeling pretty stupid that RAM Divider is one thing that i'm unable to understand :confused:

 

I site the Centon example because its 1:1 and hence i guess it would be easier to understand. I have really no idea HOW the 180 MHz figure was reached for the GEIL ONE BH5.

 

Please enlighten me :)

Share this post


Link to post
Share on other sites

200ddr is 1:1 ratio so if the FSB is 275 then the ram is also 275.

The divider works like this..

 

CPU SPEED '/. Divider = Ram speed

So for example 300FSB with a 9X multi using the 180 divider..

 

2700mhz '/. 10 (180 divider) = 270Mhz "ram speed" (DDR540)

------------------------------------------------------------

If it were the 200 divider then it would be like this

 

2700mhz '/. 9 (200 divider) = 300mhz "ram speed" (DDR600)

 

Hope this helps...

Share this post


Link to post
Share on other sites

Thank you 27floyd.

 

I'm still unclear with the logic. So, i'll try to figure it out myself and ask you guys if i'm right :)

 

ratio_how.jpg

 

Am i right in calculating the above?

 

If yes. Then i quote angry who gives some details regarding the project in his first post.

 

I'll leave this stuck to the top for a while for you guys to discuss and debunk what I have done, while I continue testing this rig...and when I am done maxing it out, I will then drop back to 250x10 and do straight 1:1 comparing since I know the BH-5 will do 250x10 1:1 easily (already Primed it at those speeds, and it is doing 250Mhz on the DDR180 divider at 275x10). The TCCD will of course do it...the differences then will be purely in what the different timings will net you (which won't be much difference either way, again making timings a moot issue unless you just HAVE to have the best benchmarks on the planet).

 

So, how do i get 250 MHz from the 180 divider formula?

The formula gives 247.5 MHz right? :O

 

 

I mean, i get the point. I concur with angry. But its just that i want to get my logic right and then go ahead and happily start experimenting and testing.

 

 

.

Share this post


Link to post
Share on other sites
Guest LookBackX2

I would say your right on target, and to really get a grasp of the concept.....start playing around in your bios.

Share this post


Link to post
Share on other sites

Thank you for the encouragement.

 

Given that even a few MHz can make a huge difference b/w a stable and unstable system. I do not think that angry rounded off the 247.5 figure to 250.

 

Now, this is what made me post the calculation and ask for clarification.

 

So, could anybody please tell me whether it was a 'rounded off' figure or what is wrong with my calculation?

Share this post


Link to post
Share on other sites

ok.

no where in your calculations did i see you use the "CPU SPEED"

This is why your calculations are WRONG.

 

It would look like this..

 

275x10=2750MHZ (cpu speed)

 

2750 (cpu speed) '/, 11 (180 divider) = 250MHZ "DDR500"

 

the 200 divider or 1:1 is always the CPU multiplyer (eg. 10 in the above case)

Each step down is an increase in "1" of this number. So the 180 divider is 10+1=11..And so on..... Get it?

Share this post


Link to post
Share on other sites
So, could anybody please tell me whether it was a 'rounded off' figure or what is wrong with my calculation?

 

Like 27floyd was getting at, the divider names (2/3, 5/6, 9/10) are in fact approximations themselves. The real divider is applied to the total cpu clock.

 

The BIOS dividers could be renamed, except the total cpu divider (/10, /11, /12) changes depending on the cpu multiplier.

 

ie. At stock, the 1:1-200 MHz divider is equal to the cpu multiplier.

 

I hope that made sense...

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×