Another PC4000 comes out... Vitesta PC4000 By A-Data

Anyone have tested them??
Use A-Data PC4000 PCB but Red, with A-Data 4ns chips and a Passive Red Heatsink
i have tested, but on my AMD setup and them seems quite better than A-Data PC4000 (that it's the best price/performance PC4000 in the market for now)
tryed on NF7-S V2.0, Bios 10, Cpu Int Enable, DC
stability tested with Prime95 & 3DMark2k1


FSB-----2-3-3-5----2.5-3-3-5-----2-3-4-5------2-4-4-5
200--------2.65v--------2.65v---------2.65v----------2.65v
205--------2.65v--------2.65v---------2.65v----------2.65v
210--------2.65v--------2.65v---------2.65v----------2.65v
215--------2.75v--------2.75v---------2.65v----------2.65v
220--------2.85v--------2.85v---------2.75v----------2.65v
225--------2.95v--------2.95v---------2.85v----------2.65v
230--------3.10v--------3.10v---------2.95v----------2.65v
236--------N/A-----------N/A-----------3.02v----------2.65v
240--------N/A-----------N/A-----------3.10v----------2.65v
246--------N/A-----------N/A-----------3.20v----------2.65v
250--------N/A-----------N/A-----------3.25v----------2.65v

at 250mhz the system performs 3760MB/s on sandramem, that it's a very hight bandwith for 2-3-4-5 timings, don't you think??
seem that this memory performance isn't affect heavy by timings as other memory type

from my test are a lot better than standard A-Data PC4000, that in my setup (same bios, all same) can't do for example over 220mhz with 2-3-3-5 at 3.2v!!! this can do 2-3-3-5 at 220mhz at 2.85v!!!
but can be that my standard A-Data are unlucky and this very lucky..

EDIT: added ----- in the table becasue it's impossible to read without, the white spaces aren't accepted...

nice scores.
and this is easyer to read:

So where did you get this ram at?

http://www.pc-memory-upgrade.co.uk/memory/ddr-ram-pc4000.asp

u can find em there

Snap, I got some of those too, from the same place as well.

Nice table, that will save me some time, thanx!

I run them 1:1 @FSB 268(intel system) at 2.75V, but that is all the voltage that my mobo has to offer so can't go beyond that.
I've seen others reporting fsb above 300 but that was with 3.3V or somthing, so I can't get there.

I noticed something very strange though.
This is kind of a long story but you don't HAVE to read it .
#######################################
I used to be a devout 1:1 man.

After reading a lot and asking some questions I came to respect the view of people who preached the gospel of the 5:4 divider. Since you've more experience than me and benchmarks to back it up I accepted that 5:4 and very good timings might be better in some (ok most) situations.

But deep down I had not changed, so I ordered som ADATA PC4000 sticks to check it out for myself.

I wanted to hit 6GB/s (Sandra memory bandwidth) and I had calculated I would get there FSB@270 (1:1) or just after.

Problem is the memory would not get there even with 3-4-4-8 timings. I set VMEM to it's (pitiful) maximum of 2.75. But the memory still gave up at 269 and bench of ~5800MB/s.

Someone wrote that increasing the chipset voltage might help too, my bios only has CPU,DDR and AGP Voltage so I upped the CPU from 1.5 to 1.7 to see if it would help. It didn't, so I gave up on the memory. But decided to see what my CPU could do at 1.7V, changed the memory divider to 5:4, left the timings at 3-4-4-8 since I did not care about memory now, FSB@288 looked good. Then just for fun I did a memory benchmark, not expecting anything much....

6054 MB/s at FSB 288 5:4 and 3-4-4-8 timings.

Seems 5:4 beats 1:1 not only with good timings but also with really crap ones Go figure...
#########################################
Br
Hepp

I should have been more specific.

Anyone know where to get them in the US? I know Komusa has some Adata, but not the ones mentioned here.

howether this Vitesta PC4000, like A-Data PC4000 from my test clocks a lot better with timigns 3-4-4-5 than 3-4-4-8, and probably 2.5-4-4-5 clocks the same as 3-4-4-5 (but under intel system this isn't a great difference)

Hey Dirty Punk, 250 fsb at 2.65 volts is insane, even at 2-4-4-5. You can't argue with the buffered Sandra scores. You will probably get better 3DMark results at 2-4-4-7+.

Are you running twin 256 or 512 MB Vitesta sticks? I imagine 256 MB.

And are you running an unmodded NF7-S???

Hepp, the buffered Sandra marks depend a lot on the board. Some boards like the Asus and Abit boards get real high buffered. The unbuffered scores are more closely tied to 3DMark performance.

I picked up two 512 MB Vitesta sticks myself, to replace my SS ADATA PC4000 sticks. Haven't tested them yet but they look promising! Your results are interesting as I hadn't planned on using them on AMD boards.

Clevor,

Seems my board prefers the 5:4 so I think I'll RMA these in favour of some good old BH5.

Are these really DS? They didn't look it, but it was kind of hard to tell with the heatspreaders on and I was in a hurry to try them so did not look all that closely. Their behaviour and poor unbuffered performance suggests that they are SS.

I tried them at FSB 260 1:1 (3-4-4-8) and 5:4 (2-3-3-7)and the 3DM01 score is the same.

But with 5:4 I can up the FSB to 280 and then the 3DM01 is certainly not the same

I think this is the wrong memory for my system.

How did yours work for you?

Anyone here have some A-Data and pushed 3.2volts?? I'm debating on getting A-data PC4000 2x512MB to replace my Buffalo PC3700. Going from 5/4 to 1/1

With all the PC4000, 256 MB are SS and 512 MB are DS.

Seems my board prefers the 5:4 so I think I'll RMA these in favour of some good old BH5.

Click to expand...

The 2.40C does run high fsb, but what board do you have? A lot of guys are rushing out to get the BH-5 stuff surfacing now, but you need more than 2.8-3.2 volts to get the most out of it. There is only one motherboard available now that can do that, the Epox. I tried two IC7-MaxIIIs and on both the VDIMM did not work over 2.8. On other boards, you'd have to do a volt mod.

Anyone here have some A-Data and pushed 3.2volts?? I'm debating on getting A-data PC4000 2x512MB to replace my Buffalo PC3700. Going from 5/4 to 1/1

Click to expand...

Some guys have gotten close to 300, 1:1 on the ADATA, but needed 3.2 volts on SS sticks. I still keep my two SS sticks of ADATA as I've gotten up to 282, 1:1 at 2.9 volts on them, CPU limited. Epox board, of course. I'm gonna test the Vitesta when I get the Epox board back up.

VDIMM mods on motherboards arn't that difficult, beyond the fact that the first one is somewhat scary.

Clevor, interesting to see your results with 4pca3+

I can't say enough about the 4PCA3+.

Case in point: I reached my highest 3D stable fsb last night: 289 with an MO step 2.40C on the DFI 875P Infinity board. I got my 9800 Pro to pass all 3D there. So I did an all out bench with the card maxed out at 428/355 (crappy, but I have no luck with ATI cards).

So at 289 fsb (3.47 gig), 3:2, DDR385, 2-3-3-6, 2.7 VDIMM I got a 20,007 score, disappointing.

On the 4PCA3+ at just 277 (3.32 gig), 5:4, DDR443, 2-2-2-5 and 2.9 VDIMM, I got a 20,167 score. It's because I get 3300 unbuffered here compared to 2750 unbuffered on the DFI board, plus the faster CAS settings. I also got identical SuperPI 1M times of 40 sec.

So you don't have to run high fsb and big CPUs for the best performance! It's too bad I cannot get above 277-279 fsb 3D stable on the Epox board, no matter what. But I'll take it.

Since the price came down, I got a 3.0C over the weekend. Mainly because I could check the step: L3337xx Malay, so I went for it. I'm looking for 240-250, 1:1, 2-2-2-5 with my XMS3500 on the 4PCA3+. That will also put me at 3.6-3.75 gig . Dare I say 3600 unbuffered at CAS 2-2-2-5?

Thanks Clevor,

That really cleared things up for me.
It was clear and to the point, just the kind of advice we all need.

I guess I already knew it, but I had my blinkers on since I wanted to believe I could get great performance out of my system, if only I could get the right memory. But you are right for a great unbuffered score I'd have to get a serious mobo and some better cooling.

Do you think I'd get better result with 2.75V using a Mushkin or Corsair PC4000 or should i just stop dreaming and accept that this system will never fly?

br
hepp

Clevor said:

I can't say enough about the 4PCA3+.

Case in point: I reached my highest 3D stable fsb last night: 289 with an MO step 2.40C on the DFI 875P Infinity board. I got my 9800 Pro to pass all 3D there. So I did an all out bench with the card maxed out at 428/355 (crappy, but I have no luck with ATI cards).

So at 289 fsb (3.47 gig), 3:2, DDR385, 2-3-3-6, 2.7 VDIMM I got a 20,007 score, disappointing.

On the 4PCA3+ at just 277 (3.32 gig), 5:4, DDR443, 2-2-2-5 and 2.9 VDIMM, I got a 20,167 score. It's because I get 3300 unbuffered here compared to 2750 unbuffered on the DFI board, plus the faster CAS settings. I also got identical SuperPI 1M times of 40 sec.

So you don't have to run high fsb and big CPUs for the best performance! It's too bad I cannot get above 277-279 fsb 3D stable on the Epox board, no matter what. But I'll take it.

Since the price came down, I got a 3.0C over the weekend. Mainly because I could check the step: L3337xx Malay, so I went for it. I'm looking for 240-250, 1:1, 2-2-2-5 with my XMS3500 on the 4PCA3+. That will also put me at 3.6-3.75 gig . Dare I say 3600 unbuffered at CAS 2-2-2-5?

Click to expand...

interesting indeed my 4PCA3+ is on folding duty right now with XMS3500C2 hehe

got some nice OCZ PC3500 Gold and Platinuum dual channel sticks to test on 4PCA3+ while my P4C800-E is being RMA

Do you think I'd get better result with 2.75V using a Mushkin or Corsair PC4000 or should i just stop dreaming and accept that this system will never fly?

Click to expand...

I don't think you should be overly concerned with buffered memory scores, as they vary a lot from board to board. I'm surprised you got higher buffered with 5:4. You should aim for the best unbuffered, paired with as fast CAS timings as you can.

The Intel boards don't have a reputation as being good overclockers, but few people have tried them. At 280 or so, 5:4 is around DDR448. You could probably run 2.5-4-3-7 there with PC4000. With DS sticks, I got 3100 unbuffered at 281, 5:4 at those settings. That is pretty darn good!

Another thing you could try is run 288, 3:2 and run 2-3-3-6 on the PC4000. With BH-5, you could probably run 2-2-2-5 since you are only at around DDR380.

Generally, the setup with better unbuffered scores will give you better 3DMark scores.

got some nice OCZ PC3500 Gold and Platinuum dual channel sticks to test on 4PCA3+

Click to expand...

Hmmm, credit card not maxed out yet Eva?

the review with the photo on amd system

nf7 vddr mod, vdd doesn't work right now
the ram runs a lot better in o/c at 2-3-3-5 for example than 2-3-3-7 or 2-3-3-11... for that reason i use 2-3-3-5
dual channel, 256mb*2, cpu int enable

Clevor,

Thanks for putting me straight again.

I just realised that my unbuffered score of 2800 fsb 260 1:1 was not that bad for a Springdale.

I was comparing it to scores produced with Canterwood and double sided sticks.

Br
hepp

I just realised that my unbuffered score of 2800 fsb 260 1:1 was not that bad for a Springdale.

Click to expand...

I just finished playing with a P4P800-D last night. And this is an early Springdale board which probably has a Canterwood chip in it, albeit castrated by the BIOS. It falls short of the P4C800-E in every category.

I did some 3DMark2001 benches last night. I am getting lower scores at 5:4, 2-3-2-5 than on my DFI 875P board at 3:2, 2-3-3-6, around same fsb (285 or so). My 4PCA3+, which can't run above 279 fsb but can run BH-5 at 5:4, 2-2-2-5 beats the scores of both boards. The 4PCA3+ is also a Canterwood board.

If you have to beg, borrow, or steal, scrape up enough money to get a CW board. The best performing boards I've seen are the Asus, Epox, and DFI. These boards have PAT on at all mem ratios. The Abit has too many issues to compete with the others (I've tried an IS7 and two MaxIIIs).