"Venice" Athlon 64 power & heat review

halcyon · Post by **halcyon** » Tue Jun 07, 2005 3:54 am

Venice is _marginally_ faster than Winchester. Usually 1-2% if at all. Not more than 5% in any scenario (clock for clock).

scruzbeachbum · Post by **scruzbeachbum** » Tue Jun 07, 2005 7:27 am

More on Venice and strained silicon.

http://techreport.com/reviews/2005q2/at ... dex.x?pg=1

jojo4u · Post by **jojo4u** » Tue Jun 07, 2005 10:15 am

http://www.hardtecs4u.com/reviews/2005/ ... index6.php

Shows that the Venice needs less power than the Winchester. Both in load and non-Cool'n'Quiet idle.

ddrueding1 · Post by **ddrueding1** » Tue Jun 07, 2005 11:24 pm

Here's my experience underclocking my Venice-core CPU.

This example is a 3500+, whose maximum multiplier is 11 and FSB is 200Mhz (therefore 2.2Ghz CPU)

I've been playing with CrystalCPUID to see what the processor can do and have made some really neat discoveries. First thing I did was OC the FSB to 228 to give a max clock of 2.5Ghz+.

1. At it's minimum voltage of 0.8v it produces virtually no heat at it's minimum multiplier (4x == 900Mhz+)
2. Still at it's minimum voltage (0.8v) it runs Prime95 100% stable at a 9x multiplier (2Ghz+)
3. It can run at it's stock clock speed (2.2Ghz) at 1v (default is 1.4v)

The thing that really blew me away was the 2Ghz+ clock speed running at only 0.8v. I'm not sure how much power this thing is actually drawing, but it is doing quite well completely passively! I'm quite thrilled.

The I · Post by **The I** » Wed Jun 08, 2005 12:08 am

ddrueding1 wrote:Here's my experience underclocking my Venice-core CPU.

This example is a 3500+, whose maximum multiplier is 11 and FSB is 200Mhz (therefore 2.2Ghz CPU)

I've been playing with CrystalCPUID to see what the processor can do and have made some really neat discoveries. First thing I did was OC the FSB to 228 to give a max clock of 2.5Ghz+.

1. At it's minimum voltage of 0.8v it produces virtually no heat at it's minimum multiplier (4x == 900Mhz+)
2. Still at it's minimum voltage (0.8v) it runs Prime95 100% stable at a 9x multiplier (2Ghz+)
3. It can run at it's stock clock speed (2.2Ghz) at 1v (default is 1.4v)

The thing that really blew me away was the 2Ghz+ clock speed running at only 0.8v. I'm not sure how much power this thing is actually drawing, but it is doing quite well completely passively! I'm quite thrilled.

What case do you use ddrueding1? - your setup looks a lot of what I'd go for, so I'm kind of curious.

ddrueding1 · Post by **ddrueding1** » Wed Jun 08, 2005 8:15 am

The I wrote:What case do you use ddrueding1? - your setup looks a lot of what I'd go for, so I'm kind of curious.

I'm waiting on a black P180, so in the meantime it's sitting on a shelf in my desk, cables everywhere. I chose to go with the Scythe Ninja over my old Scythe NCU-2005 because of the diagonal airflow that will be caused by the 2 exhaust fans.

halcyon · Post by **halcyon** » Wed Jun 08, 2005 8:46 am

Wow!

Impressive results. You have a very good specimen.

Can you try to run the following also:

http://testmem.nm.ru/snm.zip

Disable safe mode and put 100% load (setting in S&M).

It will load your system much more than prime95.

I'd like to know how your idle/load temps vary at 0.8V.

That is, if you have the time/willingness to test this.

Do note that the test in question can also crash your system, even when Prime95 does not, especially if you run all the tests (also memory and disable all safe modes in it).

regards,
halcyon

The I · Post by **The I** » Wed Jun 08, 2005 9:30 am

ddrueding1 wrote:
The I wrote:What case do you use ddrueding1? - your setup looks a lot of what I'd go for, so I'm kind of curious.
I'm waiting on a black P180, so in the meantime it's sitting on a shelf in my desk, cables everywhere. I chose to go with the Scythe Ninja over my old Scythe NCU-2005 because of the diagonal airflow that will be caused by the 2 exhaust fans.

Does this mean you're actually running it all passively, I mean, not just 'no fan on the heat sink, but a 120 mm at the back of the case' but just by convection or what?

I'm considering the p180 myself, though I think I'm going to end up going with the Lian Li V1100B for aesthetical reasons.

ddrueding1 · Post by **ddrueding1** » Wed Jun 08, 2005 10:18 am

The I wrote:Does this mean you're actually running it all passively, I mean, not just 'no fan on the heat sink, but a 120 mm at the back of the case' but just by convection or what?

There is a Nexus 120mm @ 7v on another shelf blowing across my hard drives; some circulation is likely occuring, but nothing that can be felt.

The I · Post by **The I** » Wed Jun 08, 2005 10:33 am

ddrueding1 wrote:
The I wrote:Does this mean you're actually running it all passively, I mean, not just 'no fan on the heat sink, but a 120 mm at the back of the case' but just by convection or what?
There is a Nexus 120mm @ 7v on another shelf blowing across my hard drives; some circulation is likely occuring, but nothing that can be felt.

It still reasures you a little that it's 'cool' with that relatively small amount fo air, as far as I know the Lian Li case I'm looking at has some airflow problems, but it seems to me that with the right processor and heatsink I should be able to manage it.

Mats · Post by **Mats** » Wed Jun 08, 2005 11:15 am

ddrueding1 wrote:First thing I did was OC the FSB to 228 to give a max clock of 2.5Ghz+.

I don't know which specific Gigabyte mobo you're using, but you realize that you probably get much higher speed than that? You would probably need more airflow, and I'm not saying that you actually need more speed. 2.5 Ghz max speed makes it an average or lower than average overclocker, but, 2.5 GHz at 1.2 V tells me that this one is a very good overclocker and can probably go beyond 2.7 GHz with decent air cooling.

ddrueding1 · Post by **ddrueding1** » Wed Jun 08, 2005 10:01 pm

Mats wrote:
ddrueding1 wrote:First thing I did was OC the FSB to 228 to give a max clock of 2.5Ghz+.
I don't know which specific Gigabyte mobo you're using, but you realize that you probably get much higher speed than that? You would probably need more airflow, and I'm not saying that you actually need more speed. 2.5 Ghz max speed makes it an average or lower than average overclocker, but, 2.5 GHz at 1.2 V tells me that this one is a very good overclocker and can probably go beyond 2.7 GHz with decent air cooling.

I know, but I'm not a speed freak anymore. The most CPU-intensive thing I do is encode XviD from DVD; and the 33% reduction in time from my 3000+ is plenty, saving another 5% is a non-issue for me. Cool and quiet running is my primary objective, I only OC'd it because it was "free" accoustically speaking.

Mats · Post by **Mats** » Thu Jun 09, 2005 11:38 am

The goods thing with dynamic overclocking is that you can still be a speed freak if you want to, but still have a low power system when you don't need that performance. Without that I wouldn't be very interested in overclocking at all with a new system.

ddrueding1 · Post by **ddrueding1** » Thu Jun 09, 2005 1:36 pm

Mats wrote:The goods thing with dynamic overclocking is that you can still be a speed freak if you want to, but still have a low power system when you don't need that performance. Without that I wouldn't be very interested in overclocking at all with a new system.

To effectively OC an A64 rig, I'd need to read up on it some more. This is the fastest I could get the FSB with my current RAM without increasing the voltage on anything (Hypertransport, RAM, PciE).

Mats · Post by **Mats** » Fri Jun 10, 2005 12:27 pm

Try a memory divider and see how far you can get, just for fun!
Don't forget lowering the HTT multiplier.

mb2 · Post by **mb2** » Fri Jun 10, 2005 2:30 pm

ddrueding1 wrote:
The I wrote:Does this mean you're actually running it all passively, I mean, not just 'no fan on the heat sink, but a 120 mm at the back of the case' but just by convection or what?
There is a Nexus 120mm @ 7v on another shelf blowing across my hard drives; some circulation is likely occuring, but nothing that can be felt.

what temps do u get in this situation? and any chance u could try it with completely 0 airflow across the cpu @ 2ghz? (making sure psu fan was not near cpu either)

dan · Post by **dan** » Fri Jun 10, 2005 3:30 pm

would the venice be more undervoltable than a winchestor

ddrueding1 · Post by **ddrueding1** » Fri Jun 10, 2005 3:35 pm

dan wrote:would the venice be more undervoltable than a winchestor

This I can certainly confirm. My winchestor was stable at 800Mhz requiring 0.85 volts, and stable at 1.6Ghz at 1.2v. This chip does much better.

Unfortunatly, the latest version of speedfan locks up the entire machine when it launches...still trying to figure that out.

Slidax · Post by **Slidax** » Mon Jun 13, 2005 1:41 am

Ok i recently got one of these Venice CPUs and the computer has been on all day.

From Asus probe - Cool n Quiet and Q-fan are enabled.
CPU temp - 31C
Stock fan RPM - 721
Vcore - 1.104 (due to CnQ)

This isn't under full load but even then it doesnt get too hot.

Mats · Post by **Mats** » Mon Jun 13, 2005 2:47 am

Slidax wrote:Ok i recently got one of these Venice CPUs and the computer has been on all day.

From Asus probe - Cool n Quiet and Q-fan are enabled.
CPU temp - 31C
Stock fan RPM - 721
Vcore - 1.104 (due to CnQ)

This isn't under full load but even then it doesnt get too hot.

You didn't mention the speed?

Slidax · Post by **Slidax** » Mon Jun 13, 2005 6:05 am

3500+ so 2.2ghz

CnQ knocks it down to 1ghz when idle

and by idle i mean webrowsing, copying a few files, listening to music, watching tv. that sorta stuff

i may post the results of it under load when i feel like playing games or something

edit: played BF2 demo for 2 hours
CPU - 40C
fan rpm - 1312rpm

rpsgc · Post by **rpsgc** » Fri Jun 17, 2005 9:07 am

So, do you think that the new FX-55 and FX-57 on the San Diego core are going to consume less than the current FX-55? Since the old one has a TDP of 103W and the new ones have one of "only" 89W

Because we can't really compare a 89W 3800+ Venice with a 89W San Diego FX-57, or can we?

astrayan · Post by **astrayan** » Fri Jun 17, 2005 6:13 pm

Ah, so you can get a Venice below 1.1V with an nForce4 chipset.

Why won't my A8V K8T800 go down that low?

I don;'t know if it will make much difference if its only supposed take 15W @ 1.1V and full load. The specs for these chips say that in HLT, they should take 2-3W.

perplex · Post by **perplex** » Sun Jun 19, 2005 6:34 am

ddrueding1 wrote: The thing that really blew me away was the 2Ghz+ clock speed running at only 0.8v. I'm not sure how much power this thing is actually drawing, but it is doing quite well completely passively! I'm quite thrilled.

jesus christ, 2GHz @ 0.8V - that sounds incredible. have you tried passive with that? i'm sure a xp-120 could passively cool that

Pétur · Post by **Pétur** » Sun Jun 19, 2005 7:55 am

astrayan wrote:Ah, so you can get a Venice below 1.1V with an nForce4 chipset.

Why won't my A8V K8T800 go down that low?

Same problem here, can't get my (brand new) Venice to go below 1.1V on my Gigabyte K8NXP-9

I can select it, but it limits it to 1.1V

I've also noticed that when I select a lower voltage in the bios, CrystalCPUID can't change the voltage anymore...

Peter

dan · Post by **dan** » Sun Jun 19, 2005 9:09 am

i can undervolt my winchestor-core sempron to 1.008volts at 1.6ghz. should i ebay it for a venice core?

perplex · Post by **perplex** » Sun Jun 19, 2005 10:21 am

dan do you know how much power its consuming? i wonder if Venice based Semprons are released

Mats · Post by **Mats** » Sun Jun 19, 2005 10:48 am

perplex wrote:dan do you know how much power its consuming? i wonder if Venice based Semprons are released

Yes they are, with E3 stepping and soon with E6.

perplex wrote:
ddrueding1 wrote: The thing that really blew me away was the 2Ghz+ clock speed running at only 0.8v. I'm not sure how much power this thing is actually drawing, but it is doing quite well completely passively! I'm quite thrilled.
jesus christ, 2GHz @ 0.8V - that sounds incredible. have you tried passive with that? i'm sure a xp-120 could passively cool that

dan · Post by **dan** » Sun Jun 19, 2005 10:56 am

hi
Sisandra reports that the cpu is putting out 23 watts of electricity. It is D0 stepping. i cannot help but wonder if i sell my cpu and get a venice core, i can more deeply undervolt. i do not know if it is prime95 stable.

if i can post pics i'd be happy to do so.

Mats · Post by **Mats** » Sun Jun 19, 2005 11:00 am

dan wrote:hi
Sisandra reports that the cpu is putting out 23 watts of electricity. It is D0 stepping. i cannot help but wonder if i sell my cpu and get a venice core, i can more deeply undervolt.

Why would you? Isn't it good as it is?