Sandra 2007 bandwidth results - AM2 and Intel Core 2 Duo

Since we have a lot of different platforms, CPUs and OCing modules out there right now it might be interesting to see how they compare in memory bandwidth tests and which config actually gives the highest results.

So lets see some "old-skool" SiSoft Sandra 2007 bandwidth numbers - and why not Everest scores too! :)

Please post at least the following details:

CPU+clock+multiplier (FX62, 3.2GHz, x16)
FSB+RAM clocks (200MHz, 400MHz/DDR800)
RAM type+size+Timings (PC8000, 2x1GB,tCL, tRCD, tRP, tRAS - CL3-3-3-9)
Chipset+MB model

I'll start with couple AM2 DDR800 results - buffered (Default) and unbuffered scores:

FX62, 3.2GHz, x16
FSB 201MHz, RAM 402MHz
PC8000, 2x1GB, CL3-3-3-9-1T
ATI RD580 AM2 ref. board

Sandra 2007:
buffered: 10008/9943 MB/s
http://www.xtremesystems.org/forums/...0&d=1158421860
unbuffered: 6261/6348 MB/s
http://www.xtremesystems.org/forums/...1&d=1158421860
Ok lets see some Overclocked results too ;)

NOTE: Sandra should be ran at boot-up clocks in order to see the correct result (don't adjust multiplier/FSB w/ Clockgen etc before running the test), also you shouldn't adjust the CPU multiplier in Windows if you wish to see proper Everst score.
Everest version before 3.5 give incorrect (=too high) result with 2M Allendale chips - please use the latest Everest version with 2M Allendale chips.

The settings for running unbuffered Sandra can be seen HERE

400 @ 1T, nice

No timing adjustments on this Aw9D-Max atm, about 1k behind you in buff but you take a huge lead on unbuff.

Good to see you putting up some numbers Sami.

Macci! Glad to see you're still around :)

http://afireinside.evilspork.net/AFI...-bandwidth.JPG

3800+ x2. 3812, 10x
htt 281, ram 562.5
4-4-4-8 2T
Foxconn C51XEM2AA NF590
4*512 PC5300 ballistix 4 way interleaving

Thanks for the input so far guys :up:

Lets see some more results!
I wonder what are the highest Conroe bandwidth numbers (sandra and/or everest Read+Latency)?

E6300, 3.08GHz, x7
FSB 440MHz, RAM 550MHz Vdimm = 2.45v
PC26400, 2x1GB, CL4-4-4-12-2T G.skill HZ (micron D9GMH)
Intel P965 (Rev C1) Asus P5B-deluxe Bios 0614 (no mods)

Note: all settings done via bios.

http://www.users.on.net/~bugeyes/ind...550%20copy.jpg
http://www.users.on.net/~bugeyes/ind..._02%20copy.jpg
http://www.users.on.net/~bugeyes/ind..._01%20copy.jpg
http://www.users.on.net/~bugeyes/ind..._02%20copy.jpg
http://www.users.on.net/~bugeyes/ind..._03%20copy.jpg
http://www.users.on.net/~bugeyes/ind..._04%20copy.jpg

Some quick 1:1 tests with me new set. Odd results though. Boot at 450 and CG up to 475 and vice versa. Same timings.

8x450 bios = 7558
http://img81.imageshack.us/img81/8870/450biosmv9.th.jpg

8x450 bios CG up to 475 = 7576
http://img81.imageshack.us/img81/337...75cgaj0.th.jpg

8x475 bios = 8032
http://img81.imageshack.us/img81/884/475biosdh7.th.jpg

8x475 bios CG down to 450 = 8041
http://img81.imageshack.us/img81/288...g450hh8.th.jpg

Managed to push the CPU to a bit higher clock with some chilled water (water temp +2...+8C).

system:
FX62, 3.33GHz, x12 (Vcore 1.50V)
FSB 278MHz, RAM 556MHz
PC8000, 2x1GB, CL4-4-3-12-2T (or CL4-4-4), Vdimm 2.45V
ATI RD580 AM2 ref. board
chilled water

Sandra 2007:
buffered = 12126 / 12129 MB/s
http://img.techpowerup.org/060920/sa...12_buf_12k.gif

unbuffered = 7160 / 7169 MB/s
http://img.techpowerup.org/060920/sa...2_unbuf_7k.gif

Everest 3.01 Ultimate:

Read:
http://img.techpowerup.org/060920/ev...8x12_10850.gif

Write:
http://img.techpowerup.org/060920/ev_write_278x12.gif

Copy:
http://img.techpowerup.org/060920/ev...8x12_11060.gif

Latency:
http://img.techpowerup.org/060920/ev...78x12_35.4.gif

All runs done with boot-up clocks.

loc.o, that's because Sandra doesnt detect the changes you made with CG - it will show the boot up performance. Everest works better with CG (just dont adjust the multiplier in Windows).

Thanks for the results so far - keep em coming! :)

asus p5b deluxe
e6300 allendale
g.skill 6400 c3 2gb

http://82.173.172.10/Rol-Co/bbss4-4-4-8-565.PNG

Thanks Rol-Co - great run!
over 9k with Core 2 Duo :up:

If you have the time&interest it would be interesting to see the unbuffered Sandra run and some Everest numbers (Latency in particular) too at those high FSB + RAM clocks.

edit: hehe yea its not a Conroe =)

Quote:

---

Originally Posted by macci

Thanks Rol-Co - great run!
over 9k with Conroe :up:

If you have the time&interest it would be interesting to see the unbuffered Sandra run and some Everest numbers (Latency in particular) too at those high FSB + RAM clocks.

---

nope .......allendale :D

edit: i will test the other things too, not now its 01:30 here , have too sleep fast

Tried some more all set in bios. 2.3vdimm.

HZ@525 4:5
http://img.photobucket.com/albums/v6.../5254-59-1.jpg

HZ@510 1:1
http://img.photobucket.com/albums/v6...7x5101-1-1.jpg

Geil 6400 1:1 2.3VDIMM

Quote:

---

Originally Posted by macci

Thanks Rol-Co - great run!
over 9k with Core 2 Duo :up:

If you have the time&interest it would be interesting to see the unbuffered Sandra run and some Everest numbers (Latency in particular) too at those high FSB + RAM clocks.

edit: hehe yea its not a Conroe =)

---

ok ran some test for ya macci, with allendale :D
don't know what all to shut off in the settings for unbuffered, got two here.....

sisoft Buffered
http://82.173.172.10/Rol-Co/bbss4-4-4-8-565.PNG

sisoft unbuffered
http://82.173.172.10/Rol-Co/unbuffered.PNG

sisoft unbuffered 2
http://82.173.172.10/Rol-Co/sandraunbuf.PNG

everest banthwith
http://82.173.172.10/Rol-Co/everestbb.PNG

everest cpu queen
http://82.173.172.10/Rol-Co/everestcpu.PNG

everest latency
http://82.173.172.10/Rol-Co/everestvert.PNG

Thanks for the results again sir :up: not bad! :D

The first unbuffered score is about right - the settings for the unbuffered test are posted in my first post also.

I've got a new Sandra score here at DDR1200 - let me post it for comparison (I have a feeling it might be the highest DDR1200 Sandra score I've seen) =)
http://img.techpowerup.org/061004/sa...00%2012600.png

Quote:

---

Originally Posted by macci

I've got a new Sandra score here at DDR1200 - let me post it for comparison (I have a feeling it might be the highest DDR1200 Sandra score I've seen) =)

---

now your going to make me fire up the SLI M2R/G and put ungodly voltage through my 8500 C5s ;) . 12500 is pretty bloody incredible though, not sure if i can touch that.

this is the best i managed on the M2N32-SLI:
http://img.photobucket.com/albums/v2...0_sandra-1.gif

i can also run DDR900 1T so i will see what that spits out on Sandra for comparison when i get setup again. i'll be back...

http://i55.photobucket.com/albums/g1...500fsbband.jpg

GeIL PC2-6400 C3

http://img138.imageshack.us/img138/4...200rzf9.th.jpg

http://img101.imageshack.us/img101/850/sandract9.jpg

http://img.photobucket.com/albums/v6...andra8x555.jpg

BUMP :D

Lets see some updates for this one too!

Good idea. Some months have passed and hopefully some better results can be had. Sadly my rig doesn't even touch the results from 6 months ago. :(

Quote:

---

Originally Posted by macci

BUMP :D

Lets see some updates for this one too!

---

Havent done this benchmark since khx3200 days ( I still remember your khx3000:D )but for you Ill run some test tonight..... Very nice to see you around sir...

Quote:

---

Originally Posted by macci

BUMP :D

Lets see some updates for this one too!

---

NUMA FTW :D

http://img179.imageshack.us/img179/2549/511memser8.jpg

thats sick :toast:

can you do 300x10 with the mem at 1200 that would be insane

pls use Everest 3.50 or newer version... any earlier version has memory read result bug, gives u way too high score... not real result! pls update!

Quote:

---

Originally Posted by laragirl83

pls use Everest 3.50 or newer version... any earlier version has memory read result bug, gives u way too high score... not real result! pls update!

---

If you're referring to my screenshot, then I suggest you do a little research about NUMA before accusing me of posting a "fake" result.....and if you knew anything abou NUMA, you would know that Everest doesn't correctly identify bandwidth in NUMA-enabled computers...

@ MSI - Thaks....sorry I don't have any recent updates for the Quad Opty rig - it's been quite awhile since I last ran them...I'll update when I get a chance....

stephen, nobody has called your result a fake, you didnt even ran Everest did you??? Why did you get offended? relax! :)

laragirl was referring to Everest benchmark and Read bandwidht bug with 2MB Allendale chips. There are some 2MB allendale scores posted on this thread which show too high read bandwidth.
So yes please use the latest versions.

again please keep those results coming! :D

Quote:

---

Originally Posted by s7e9h3n

If you're referring to my screenshot, then I suggest you do a little research about NUMA before accusing me of posting a "fake" result.....and if you knew anything abou NUMA, you would know that Everest doesn't correctly identify bandwidth in NUMA-enabled computers...

@ MSI - Thaks....sorry I don't have any recent updates for the Quad Opty rig - it's been quite awhile since I last ran them...I'll update when I get a chance....

---

no offense, i wasnt talkin' about you specifically!:cool:
and I also didnt say that those benchmarks were made on purpose with earlier version, maybe some of u just didnt know about the issue.
so i just saw the screenshots and it occured to me, that most of them 3.00.xx version:)
i tought i make note:p:

anyway here is mine, with Everest v3.80.934:
http://xs314.xs.to/xs314/07171/everest_read.jpg

Some of my test

Quote:

---

can you do 300x10 with the mem at 1200 that would be insane

---

Something like this you mean? :D
http://img.techpowerup.org/070424/sandra_608.png
(btw, it is not on the ASUS 4X4 board)

It gets even better with higher CPU clock:
http://img.techpowerup.org/070424/sandra18800.png

Unbuffered is not bad either:
http://img.techpowerup.org/070424/sa...r1212unbuf.png

^^what mb was that done on???

Mine

Why are you posting processor arithmetic scores here??

E6300 440 x 7 = 3.08 GHz
Gigabyte DS3 rev 1, P965 C1
2 GB A data 8500 at 1100 MHz 5-5-5-9

http://www.xtremesystems.org/forums/...6&d=1188210594

Quote:

---

Why are you posting processor arithmetic scores here??

---

Cause I'm an idiot and posted the wrong screen shot.

msimax,

link (the 4x4 one)

The question might be not needed..but that FX74 it bugged macci?(if you did test that already)

Quote:

---

Originally Posted by Lancer33

Cause I'm an idiot and posted the wrong screen shot.

---

Hehe, that's better :D

BeardyMan, its not bugged, its NUMA! :D

Quote:

---

Originally Posted by Lancer33

Mine

---

Quote:

---

Originally Posted by LIKMARK

Why are you posting processor arithmetic scores here??

---

Quote:

---

Originally Posted by Lancer33

Cause I'm an idiot and posted the wrong screen shot.

---

lmao :rofl:

Quote:

---

Originally Posted by macci

Something like this you mean? :D
http://img.techpowerup.org/070424/sandra_608.png
(btw, it is not on the ASUS 4X4 board)

It gets even better with higher CPU clock:
http://img.techpowerup.org/070424/sandra18800.png

Unbuffered is not bad either:
http://img.techpowerup.org/070424/sa...r1212unbuf.png

---

Bloody Hell!! :eek:

Thats just F'kin Awesome bandwidth macci. Amazing results! :toast2:

Intrigued as to why the bandwidth Efficiency is so low tho! Can you explain?

Also, do ya think the next Gen of Intel Chips with their own on-board Memory Controllers will be as good?
I understood they were scheduled for 1st Quarter '08 when Intel released their Roadmap on the C2D pre-launch!

*Bump* - Did I say something!?!? :shrug:

Quote:

---

Originally Posted by Kin Hell

*Bump* - Did I say something!?!? :shrug:

---

you might want to read the thread the results are skewed, they are way off due to NUMA. i thought the same thing.lol