Venice 3000 @ 3ghz ;)

[maverik-sg1]

Okay been playing with this baby for a few days - seems the memory controller aint that strong as RAM will memtest stable for 5 hours @ 318Mhz 3-3-3-10-9-18 but wont run 1:1 with the venice past 291Mhz.

Here's the clocks (not bad for a CPU that runs 1.8ghz normally):

http://valid.x86-secret.com/show_oc?id=73512

 

[K404]

Thats not totally true mav...if you keep the RAM speed the same for both you should get the increase from the CPU speed.

Dividers arent ideal, I admit, but e.g 260x10 1:1 compared to 313x10 on a 166 divider (RAM at 260 on both) the 313x10 will be better.

Kinda extreme example coz theres a 530MHz clock difference, but your score suffered coz the RAM speed was slower.

K

 

[maverik-sg1]

Thats not totally true mav...if you keep the RAM speed the same for both you should get the increase from the CPU speed.

Dividers arent ideal, I admit, but e.g 260x10 1:1 compared to 313x10 on a 166 divider (RAM at 260 on both) the 313x10 will be better.

Kinda extreme example coz theres a 530MHz clock difference, but your score suffered coz the RAM speed was slower.

K

Okay so I ran 3dm2001 at 325x9 (2.925Ghz) with 166mhz ram setting (271Mhz ram) and scored just over 27890 marks.

Running the RAM at 200Mhz (1:1) with CPU at lower over clock speed of 291x9 (2.619ghz) and scored 30105 3dmarks

So 2.62Ghz 1:1 ratio, is faster than 2.93Ghz with a ram divider (30105 vs 27890), that's extreme as there is a 300mhz difference, yet the slower clocks running 1:1 is still considerably faster.

I think its safe to say that it is strictly true for 3d gaming that having higher clocks but using a RAM divider is a real performance hit.

Just my 2p.

 

[K404]

Controversial statement!!

We`ve been told since A64 came out that dividers made no difference..they lied to us??

i`m gonna have to test this ideas out, i`m not saying I dont believe ya, but I gotta see for myself

 

[stocky]

Hehe, I never like running dividers! I agree that it is worth investing in RAM that can match the speed and timings which run well with that of the CPU.

Anyway, as long as you have the right RAM or Multipliers you won't ever need to use dividers

 

[maverik-sg1]

Will post pics later

It's not unusual for a certain amount of 'spin' to accompany the corporates statements when they sell us a concept.

 

[maverik-sg1]

Humble Pie Time

Not sure what I was thinking before must have got me all dazed and confused and got the results back to front (stomach cold today ).

Firstly Apologies to K404 :$ as I was not correct in my statement where I said the divider with higher clocks demonstrated less performance (although at the same core speed, there would be a hit, in my feeble defense).

So as promised - benchie pics for ya with clocks/memory ran.

Mav (the humble)

 

[OV3RCLOCK3R]

Not bad for a budget cpu, Wht gfx card are you using Mav?

 

[sacha35]

very nice Maverik what motherboard is that useing

 

[maverik-sg1]

Using DFI Lanparty EXPERT and 7800GTX 256MB 550/1400

Was not pushing the graphics side really as I wanted to compare high CPU with mem diveder against best cpu with 1:1 memory.

Suicide shot - breaks 3ghz - Thats a 1.2Ghz overclock

http://valid.x86-secret.com/show_oc?id=73515

New shot:

http://valid.x86-secret.com/show_oc?id=73517

Mav

 

[sacha35]

Hi maverik, was that with phase cooling as i see you are useing a high vcore?

 

[maverik-sg1]

Yes Mate using Phase cooling - check pics on first post -19 cpu temp

Here's an 01 bench at over 3ghz :O

Mav

 

[K404]

apology totally accepted. No-ones perfect, I`ve made similar mistakes

Your results are much more like what I was expecting now

Its a pretty good feeling to break 3GHz with a x9 multi, innit?

 

[maverik-sg1]

Its a pretty good feeling to break 3GHz with a x9 multi, innit?

Actually mate yes - surprisingly, even though it's not UBER in the grand scheme of things - it was a lot of fun and quite satisfying.

Totally superpi 32m stable at 322x9 (26m 05s).

Mav

 

[sacha35]

much more satisfying getting a 1.2GHz overclock now thats value for money, if only the new FX60 could hit 3.8 GHz, i know its two core but that would be some clock.

 

[welshtom]

ahhh i remember my old venice, took it to 3.3ghz

 

[maverik-sg1]

ahhh i remember my old venice, took it to 3.3ghz

Screenies? FM linkies? CPU validation link?

Which one was it, a 3000? My overclock on this is x1.67 over stock.

Nice clock, but Dazboots got his 3800 up to 3.4 Ghz http://valid.x86-secret.com/show_oc?id=14033 but in ratio's it's stock x1.41 - the higher the stock clock the higher the overclock, but the lower the bang per buck - has to be said that the venice series are great CPU's though.

I might try getting a few more benchies under it's belt, this time with some tweakage........ just to see how close I can get to some older scores - bang per buck, phase plus venice 3000 (approx £650) is probs better than FX57 (£680) and air (and some water cooled), plus once you got the phase unit you have a sustainable overclocking system and can carry on buying cheaper CPU's (well you could make a 3800 x2 run at 2.8ghz+) and make them go faster than £680 ones (at stock of course)

Sometimes it's not about records, just getting the most out of what you have. In fact if I had not got the benching bug and was just into everyday stuff (games, movies etc...) I'd probs do that (buy the lowest clocked x2 and clock it under phase, not to stupid speeds, just that 1ghz overclock mark)

Mav

 

[SAS91]

lol yea do cus ive got one

 

[welshtom]

hmm seems thats the only pic i can find. got a few 1mb times knockign about too.

Not a quick time i know. Just 3.24ghz for very little volts really. around 1.6 if i remember. Was with my mach 2.

0512 Dpbw i think she was .

Got a cpuz link somewhere but god knows what it is lol.

Tom

 

[maverik-sg1]

Fook Tom - Thats a great clock, more so with the volts you used, congrats man!!

Superpi time was not too shabby either.

Which board was you using?

Mav