Printable View
Ok guys, This is the place to post your results of bandwidth goodness, tweaks, chipset Vmod, etc. with the 8RDA+. This is an effort to keep all the discussion unrelated to Vmods out of the Vmod forum and consolidate... :D
I'll start off with some mild bandwidth at 218, Cas2, 4-3-2. I noticed that setting the Trp to 2 made a huge difference in the Float score..
229x10, 5-2-2-2 timings. VDD @ 1.85v, VDimm @2.9v
http://www.pc-extreme.co.uk/personal/mem229-52.JPG
http://www.pc-extreme.co.uk/personal/pcmark228-52.JPG
The VDD mod was excellent, and the only way to get tight timings at high FSBs for me was to set at 6-3-3, then burn-in for a few minutes and go to 5-3-3, to 5-2-2 and I've got it to 4-2-2 just not stable enough to pifast. Setting to Aggresive gives about 20 more points too :) The PCmark is not too high because it tests video and I'm using a ti4200 @300/600.
I've matched your sandra, however, this is at 4-2-2-2 timings, VDD @ 1.97v!, Agressive CPU. Looks like going with the 4 instead of 5 for Tras is worth about 4mhz fsb.Quote:
Originally posted by Bushboy
229x10, 5-2-2-2 timings. VDD @ 1.85v, VDimm @2.9v
I knew it...The 4 Tras does help, but I know for a fact my sandra scores are lower than what they should be. Look at my float score and look at yours (100MB/s difference)...something is a bit off and I dunno what it is. Also my estimated max bandwidth is higher too, I have no clue what happened there :confused:
Dual bank bushboy. It does wonders for Float :)
Dunno if it makes much difference but that is with Sandra 2003.
..and yes, Dual Bank/Channel. CPU@~2600mhz also.. my friggin screen was flashin quite a bit @ 225, even with the 1.97Vdd. I'm thinking at this voltage I need some watercoolin on that Nbridge..
:D
Oooh I thought it was only single channel. That's very impresive 225mhz DCDDR. All this time I've been comparing float to dual channel thinking mine was off :p
Here's my current dual channel high...
(anybody else think Sandra is only testing one channel?) :confused:
Yup it is Sandra doesn't recognise dual channel memory at all yet.
227fsb but score a little lower 6/2/2/2 at 2.9v 9x227fsb
dual channel
http://www.geocities.com/winxp/8rda2...?1041579486234
226.99 x 11.5
timing: aggressive 2,2,2,7
dual channel synchro
vdd 1.87v air cooled
v-ddr 2.88v
http://membres.lycos.fr/ftpad/images/record-1e.gif
What is the deal with the ram timings on this board... At 217fsb, I went from cas2, 4-2-2, to 3-2-2, and saw no difference in sandra bandwidth... The change from 4 to 3 being the Tras.
Additionally, I see some of you running 2,2,2,7 while others are 6,2,2,7... (last two threads)..
So are you guys just quoting your timings backwards? I always quote in the order they appear in bios.. Tras, Trcd, and Trp..
Just curious as I'm trying to find the best aggressive timings right now... just dinkin around.. :D
I haven't tried going from 4 to 3, but I did notice a significant change in 5 to 4, in 3dmark too. I'm impressed with how tight I can run timings at such high FSBs...I remember my IT7 that didn't allow me to run 2-2 timings at all! And as a result...15k with a ti4200 :)
That's the only change I noticed too.. Going from 5 to 4.. 4 to 3 didn't do jack! :confused:Quote:
Originally posted by Bushboy
I haven't tried going from 4 to 3, but I did notice a significant change in 5 to 4, in 3dmark too...
does the prob still exist where you lose massive bandwith using mults under 10.0?
Yes, but for people with Tbred Bs, it shouldn't be a problem It's a real problem for lower clocked CPus though unfortunately...Again, a BIOS fix is needed.Quote:
Originally posted by Emericana
does the prob still exist where you lose massive bandwith using mults under 10.0?
Wierd
Here's a PCmark run at 220 5 2 2 2
http://mywebpages.comcast.net/ddr333/Epox/pcm06.JPG
http://service.futuremark.com/compare?pcm=972895
Sandra at those settings
http://mywebpages.comcast.net/ddr333/Epox/san220.JPG
Tried 4-2-1 timings and it wouldn't even go into bios.. :confused: must not like the 1!
I dd a little checking with my thermal probe this morning after thinking about RichBa5tard mobo dieing from chip overheating from vdd mod. Mine is set to 1.84v, I have 2-60mm fans one blowing in area of mem an one lower blowing in area of southbridge with heatsink on it. I disconnected the fans to make it easier to check temp on the 3037 an the small one just to the left of the 3037. Booted up an started prime figuring putting a load on system would give me a more accurate temp reading of chip temps. Without fans 3037 42C, small chip to left 58C, prime ran for about 10 minutes an errored this is at 10x205fsb 2.77v to mem. Yesterday had no problems running prime, so what is going on here. Started prime again an 5 minutes error again. Reconnected fans an check temp running prime again check temps 3037 32C small black chip to left 38C an prime running now for about 3hrs an no errors. So cooling this area made a diference.
Just rechecked chips an 3037 30C an chip to left 33C now.
Good work WinXP! :toast:
hello,Quote:
Originally posted by AKRedneck
What is the deal with the ram timings on this board... At 217fsb, I went from cas2, 4-2-2, to 3-2-2, and saw no difference in sandra bandwidth... The change from 4 to 3 being the Tras.
Additionally, I see some of you running 2,2,2,7 while others are 6,2,2,7... (last two threads)..
So are you guys just quoting your timings backwards? I always quote in the order they appear in bios.. Tras, Trcd, and Trp..
Just curious as I'm trying to find the best aggressive timings right now... just dinkin around.. :D
yes, it' s 7,2,2,2
t ras / t rcd / t rp / cas
PCMark sandra follows... :D
Mhh, Sandra detect only one Channel?
Why do E7205 get over 4 GB/s wit FSB lower than 200 MHz?
That is because the P4 uses Quad channel FSB, and the Athlon only DDR. I wont do the math for you, but i hope you get the pictureQuote:
Originally posted by KingInge2000
Mhh, Sandra detect only one Channel?
Why do E7205 get over 4 GB/s wit FSB lower than 200 MHz?
Sandra shows the bandwith the CPU can handle. Thats why IMHO you won't get any better result with 2 channel testing.Quote:
Originally posted by DuraN
That is because the P4 uses Quad channel FSB, and the Athlon only DDR. I wont do the math for you, but i hope you get the picture
Only the max. bus bandwith would increase!
BTW :
http://www.inge2000.de/sandra230.jpg
Where is the VDD mod posted?
Found it:
http://www.xtremesystems.org/forums/...5&pagenumber=1
The P4 can accept more bandwidth than the Athlon XP, that is what i ment. As a consequence of that, the efficiency of DC for P4 is much higher than DC for Athlon XP's
Is there any efficiency of DC 4 the XP?
Thats up to you to decide :)
but i dont see that much of a performance gain
I don't know if anyone has tried tweaking in the voltage with the vdd mod, so I wanted to see really what voltages I needed for the fsb I was running. Less voltage with the vdd mod to me means less chance of killing my parts. I was looking for a stable system with no screen flicker.
Stock mobo mem 195fsbx10.5 2.63v vdd 1.58v
Run prime, no screen flicker
10x205fsb 2.77v vdd 1.58v
Will not run prime, screen flicker
10x205fsb mem 2.77v vdd 1.70v didn't go lower
run prime for an hour, no screen flicker
9.5x215fsb 2.9v 1.70vdd
prime error, screen flicker
9.5x215fsb mem 2.9v vdd 1.79v
Prime running about 2hrs., no screen flicker
I would try tweaking in my voltage, you may not need as much as you maybe running for a stable system.
Used 2 sticks of memory dual channel , mem timing
expert,aggressive,100%, 6/2/2/2
Same opinion! It was a not-2-b-answered questionQuote:
Originally posted by DuraN
but i dont see that much of a performance gain
;) or what u guys call this in english.
What's about that screen flickering? I believe it's a kind of overheat. Maybe also the mosfets? I can run Prime @ 224 but only 221 is workable because of this f***ing flickering.
Its a retorical question i think :)
dc gives me another 50mb'S in sandra membench.
Al ittle bit of improvement...burning in goes a long way :)
http://www.pc-extreme.co.uk/personal/mem231-5.JPG