DDR490 TwinMos - 2.6C ***WOW*** (P4G8X vs P4C800) SpecViewPerf 7.0 Benched

[mechBgon]

I do actually do large WinZip compression/decompressions once in a while at work... the other day I compressed 945Mb of Office2000 Disc 2 Administrative Installation Point so I could fit it onto a CD-R for backup, for instance. Times like those, I'm glad I have 15000rpm SCSI on my side but I could definitely use a 2800+ too!

Yeah, no offense meant, but Sandra is teh

. UT2003 Demo botmatch is something I'd pay more attention to, or Comanche 4, something that shows a system actually doing what systems do in real life to some extent.

 

[jhites]

Looks like I am going to settle in at 268fsb @ 1.675vcore

This chip ran default vcore up to 260fsb=3380Mhz=1040fsb
270-275fsb took 1.70vcore

Prime95 cut me back to current settings in my sig.
Still will be testing stability more but for now this is
Prime95 stable.

Still having to run the Corsair TwinX-PC3200 at 3:2 ratio
2-2-2-6 as it would not run 5:4 above DDR416 at any
timing settings or vdimm.

 

[Duvie]

Well if these p4c chips are anything like the C1 p4b chips then prime95 does not mean SHITTTTTT!!!!

definitely test it thoroughly with some intense cpu programs like encoding to see if it fails...many like Thugs, Lastride, and I showed awhile back that pirme95 ran for 6-12 hours didn't insure you could run other programs for 10minutes or more!!!

 

[jhites]

Originally posted by: Duvie
Well if these p4c chips are anything like the C1 p4b chips then prime95 does not mean SHITTTTTT!!!!

definitely test it thoroughly with some intense cpu programs like encoding to see if it fails...many like Thugs, Lastride, and I showed awhile back that pirme95 ran for 6-12 hours didn't insure you could run other programs for 10minutes or more!!!

I agree with you about the Prime stability testing and will be running other programs.
May run that Seti that MechBgon listed earlier in this thread, as well as, some
that you mentioned when my new HyperX comes in on Monday.

The main point that I come away with after hours of going through
this is that my 2.53b on P4G8X is about on par with my 2.60C on the P4C800.
The only gain may be some increased memory bandwidth and that would not
occur until above 260fsb. The negative side is that it is taking a higher vcore
on this 2.6C to get about the same performance as I had on my 2.53b.

Another point that is pretty obvious is that a C chip is a waist of $$ if someone
is going to use it on a P4G8X board.

 

[THUGSROOK]

wait til the 1mb cache chips come out

 

[jhites]

Ran that Seti Bench File that mechBgon suggested in an
earlier post but really don't have a clue as to how it compares
to other systems or where to find this information.

It shows:
cpu_time=7856.796875
2 hours 10.8 minutes

 

[Thor86]

Originally posted by: jhites
Still having to run the Corsair TwinX-PC3200 at 3:2 ratio
2-2-2-6 as it would not run 5:4 above DDR416 at any
timing settings or vdimm.

That's too bad. Oh well, thanks for the thorough review though. 268fsb is still pretty amazing though, and I can't wait until your KHX memory comes in to test!

 

[jhites]

PiFast ver4.2 and Seti Bench added for reference.

268fsb=3484Mhz=(3:2 DDR357 @ CL2-2-2-6)

I don't have anything previous to use to compare these tests,
so maybe someone else might post what they get for comparison.

 

[mechBgon]

2:10 is a great SETI result

For comparison purposes, my AthlonXP 1600+ does 0.417-angle range SETI work units in about 3:48 on an nForce 220D motherboard (old nForce, one memory controller). Overclocked to the 166/166 memory/CPU bus for 1750MHz on a KT333 board, it did the Ars SETI benchmark in about 3:15. It would be interesting to see how the P4G8X did with the SETI benchmark. Hexus benched some systems and found that an i845PE/2.53GHz setup was doing about 3:00 (linkie). Evidently dual-channel is bringing a nice performance boost to the P4 versus single-channel (duh )

Another interesting variant would be to run two instances of SETI at the same time on both the P4C800 and the P4G8X, to see how they fare in Hyperthreading performance compared to eachother.

 

[jhites]

Originally posted by: mechBgon
2:10 is a great SETI result

For comparison purposes, my AthlonXP 1600+ does 0.417-angle range SETI work units in about 3:48 on an nForce 220D motherboard (old nForce, one memory controller). Overclocked to the 166/166 memory/CPU bus for 1750MHz on a KT333 board, it did the Ars SETI benchmark in about 3:15. It would be interesting to see how the P4G8X did with the SETI benchmark. Hexus benched some systems and found that an i845PE/2.53GHz setup was doing about 3:00 (linkie). Evidently dual-channel is bringing a nice performance boost to the P4 versus single-channel (duh )

Another interesting variant would be to run two instances of SETI at the same time on both the P4C800 and the P4G8X, to see how they fare in Hyperthreading performance compared to eachother.

Unfortunately I sold my 2.53b chip and just packed it for shipping in the morning.
I do still have the P4G8X in my wifes system but it is only running a 2.4b @ 150fsb.

Maybe someone with the P4G8X and 2.53b could run it for a comparison.
Maybe Thugs, since he has a setup very close to what I had.

I will be running some of the P4C800 benches over with the HyperX PC3500
installed. Hopefully with a better mem ratio and higher bus.

 

[jhites]

Duvie

That SpecViewPerf is almost 300Mb (20 min download)
Took longer to download the program than it did
to run the bench.

 

[orion7144]

I'm very interested to see how the SETI times are if you run 2 instances of it via HT. I run follding on my pc's and since it uses 100% cpu I think it is a very good test for OC'ing stability. I think it will be interesting to see how 2 instances on an HT enabled pc will

1. affect the CPU temps
2. will it limit the OC'ing ability since it will be using 2 pipes. i.e. using 2 instances of folding will put 100% cpu utilization on both "pipes" so in theory the cpu temp should increase.
3. what about the voltage when using both pipes in an oc'd pc. Do we really know yet what the upper limit will be for Vcore?


Just some thoughts.

 

[SupermanCK]

nice nice...tonz of more real apps benchies...thx much...
i guess the HyperX3500 is better than XMS3200LL

 

[SupermanCK]

hey jhites
you might want to double check your result for "Memory Unbuffered" Test at 270FSB (432DDR)...it is a lot lower than running it at 260FSB (400DDR)

 

[jhites]

Originally posted by: SupermanCK
nice nice...tonz of more real apps benchies...thx much...
i guess the HyperX3500 is better than XMS3200LL

There's very little difference actually between running
the XMS3200 at 260fsb CL2-2-2-6-1T and the HyperX3500
at 270fsb CL2-3-3-6-1T. Only about 2% on every bench
that I retested. But it makes me feel better running DDR433.

I am going to run that Seti Bench using 2 instances to see
how it effects system stability and temps as
orion7144 mentioned. Sounds interesting I think.

 

[mechBgon]

 

[jhites]

Originally posted by: SupermanCK
hey jhites
you might want to double check your result for "Memory Unbuffered" Test at 270FSB (432DDR)...it is a lot lower than running it at 260FSB (400DDR)

I did double check it and that is what Sandra shows
with the relaxed timings for Unbuffered.
Buffered went up about 2%
Unbuffered went down about 5%
Have no explanation other than timings but ran it several times
and they all came out the same within a few points.

I have 2X512 sticks of TwinMOS PC3700 4.3ns due in
tomorrow afternoon. Be very interesting to see if I can
run it at tighter timings at DDR433 since it is DDR466.

 

[jhites]

Here are the results for running 2 instances of the Seti Bench
at the same time. Started within a couple seconds of each other.

cpu_time=10717.125000
cpu_time=10716.937500

As you can see it took almost 3 hours instead of the
2hours 10min on the first run. The temps went up to
51C for a high. My idle has been around 32-36C
depending on the ambiant in the house.

 

[mechBgon]

Very interesting! So it can do one WU in 2:10, or two WUs in about 3:00. Looks like this is a task that Hyperthreading is well-suited for. With the SETIDriver utility, it would be simple to not only run two clients at a time, but keep them off your taskbar too.

 

[SupermanCK]

I have 2X512 sticks of TwinMOS PC3700 4.3ns due in tomorrow afternoon.

where are you getting these babies??? and how much?

 

[jhites]

Originally posted by: SupermanCK

I have 2X512 sticks of TwinMOS PC3700 4.3ns due in tomorrow afternoon.

where are you getting these babies??? and how much?

MemoryX One of the links from the
where to buy at the TwinMos Website.

 

[anomaly]

I can't wait to see your results. I really hope winbond will release the CH-5/CH-4 chips before DDR II. I think we need some chips that would support 330fsb =)

 

[Thor86]

I have 2X512 sticks of TwinMOS PC3700 4.3ns due in tomorrow afternoon.

. Definitely interested in some 1:1 memory ratio scores if possible.

 

[cpars]

Originally posted by: Thor86

I have 2X512 sticks of TwinMOS PC3700 4.3ns due in tomorrow afternoon.

. Definitely interested in some 1:1 memory ratio scores if possible.

The twinmos 3700 seems pretty sweet so far, I have it at 237fsb @1:1 on a 2.8C. Looks like the cpu is maxed out around this speed.Testing different timings now

 

[jhites]

Originally posted by: Thor86

. Definitely interested in some 1:1 memory ratio scores if possible.

OK - Just because you asked
Posted a few of the 1:1 up to DDR490 @ 245FSB
I just ran it with 2.5-3-3-7 to see how high stable it would go
I posted the DDR492 but MemTest86 got errors so it is disqualified.

CPU_Z **DDR490**
CPU_Z **DDR492**