[Alexandre13]

More is less

I have added 4GB for a total of 8GB on a Vista 64 bit computer.
(4 x 2GB corsair xms2-6400-800 MHz)

time: 21m04s
Benchmark: 499.75

Slower than the result I have posted for the same computer with 2 X 2GB corsair xms2-6400-800 MHz.

time: 20m 35s
Benchmark: 514.47

Any idea

[polynurb]

More is less


Any idea

i think that's because more Ram has longer latencies.
All memory I/O has to follow a longer pipeline.. hence it makes it a little slower. it is also more difficult to overclock 8gb than 4gb of ram, afaik that 's related to the same issue.

so don't worry too much, its quite normal.

[Alexandre13]

Thanks polynurb

I discovered after i sent my message that the added memory from CORSAIR is 1.9V and the other CORSAIR modules are 1.8V.
I retested the 1.9V modules in a 4GB configuration and they are slower than the 1.8V but faster than the mixed 8GB configuration.

The computer is a Gateway FX 7026
Originaly with 2 X Samsung 2GB 667 MHz DDR2 SDRAM Memory Modules
Four memory slots support the following:

Support non-ECC DDR2-667/800
Support 1.8V DIMMs
Support 256 MB, 512 MB and 1 GB DRAM technology
8 GB maximum system memory
Support dual-channel interleave mode

[sandykoufax]

---------------------------------
benchmark

robur Q9450 / 534
I sj new Q9300 / 510
Alexandre Q9300 / 514
Andpoint Q9300 / 504

kkm Q9300 / 613 ?? why?


anyway, it seems that intel's new 45n cpu is not so dramatically fast than their 65n cpu.

[-Adrian]

I'll mail him, he was the first with a Q9300, so i had no comparison at this time. He has an odd comment to his entry ("BAD BIOS 2.0GHz"), so i'll see if can shed some light on that.

I still get a lot of odd entries, and by now i simply delete them without further investigation. My readme should explain all nono's.

[-Adrian]

The super-fast Q9300 was an error btw, so disregard that. The others are accurate.

Since 1.7 doesn't seem to show a noteworthy speed difference, i might as well keep the 1.6 list on the front page. No point in starting a new one until the core gets some major rework.

I had an idea about collapsing all entries for the same CPU (e.g. Q6600) into a single row showing the averaged result of all. One could then open the column to see all contributing entries.

Don't know if i can code that, but it would seem like a wonderful solution to the ever growing size.

[sandykoufax]

The super-fast Q9300 was an error btw, so disregard that. The others are accurate.

Then, the inte's 45nm cpu(9000series) is somewhat disappointing.

They has not much difference than 6000 series.

Anyway, collapsing idea seems good.

[JackAxe]

The main difference between the 65nm and 45nm chips, is power consumption. The Q9xxx chips achieves a similar performance with less power.

http://www.hardwarezone.com/articles/vi ... id=2&pg=11

MicroCenter is selling the Q9450 for only $279. BTW :]

<]=)

[w i l l]

Haven't posted on the Benchwell site before (just downloading scene) - I noticed that it says 'entries closed'. Why is this? How up to date are the entries listen on Benchwell.com?

[-Adrian]

The difference between 1.6 and 1.7 is like 5 seconds.

If i open a new table for 1.7 it'd be waste to not have all these entries on the frontpage. If i let users only select 1.6 (since it's basically the same), they probably wouldn't want to enter. If i mixed 1.6 and 1.7, it'd be confusing to people.

So i closed it for now. The results still hold true for 1.7, it's up to date. I have no better idea until there's some major change to the engine and thus speed.

[w i l l]

I dont think it'd be confusing. Afterall there's a 'Maxwell Version' column.

[w i l l]

My Benchwell is 2hrs and counting. Looking good. Is there a table for the worst?

[-Adrian]

Do you mean worst render times? You can sort the list by clicking on the column headers.

[w i l l]

4h 7m 44s appears to be the time to beat.

And I have....

[Bubbaloo]

What are you using to render, a calculator?