PDA

View Full Version : 2-2-2-6@5:4 or 3-4-4-8@1:1


Cho
05-24-04, 02:25 AM
I wonder, which one should performs better? :confused:

aNTiChRisT
05-24-04, 06:13 AM
Heh, its different in any PC. And specify the mhz... lol

100mhz 3-4-4-8 1:1 is worse than 200mhz 2-2-2-6 5:4 :-)

I never liked dividers, but you'll have to run some bandwidth test on it to find out.

~t0m

Suma
05-24-04, 06:15 AM
I agree with aNTiChRisT.
Just run some bandwidth test and some 3d tests too.

Suma.

NoteUser
05-24-04, 09:08 AM
Originally posted by Cho
I wonder, which one should performs better? :confused:

The 2-2-2-5 would respond faster and is very consistant compared to high bandwidth memory.

The goal is to run them tight and as high as possibly avoiding high voltage and the slow 3:2 divider.

aNTiChRisT
05-24-04, 11:04 AM
Yes, but as i said pc2100 2-2-2-5 is slower than pc4800 3-4-4-11 so until he specifies what sort of fsb he is running we can only speculate ^_^

Better yet, he could have found out himself and saved threadspace :P

~t0m

Cho
05-24-04, 12:43 PM
For example, fsb=250, then,
PC3200@5:4, 2-2-2-6, and
PC4000@1:1, 3-4-4-8
Which one will perform faster?

Btw, I can't find this out myself because I haven't built my new rig yet. I ask this so that I can have more information to choose new ram to build a new rig.

Tatuya
05-24-04, 12:54 PM
It depends on the processor and mobo but with 250 FSB I assume it's a P4. IMO the tighter timed stuff will run the best. Justification for this includes that P4s love the 5:4 ratio and high FSB. Plus if you are running it that high at tight timings that is a plus.

To reiterate (a bit), if you are running both of these @ the same speed (as you stated above) the first should be the best as it has better timings @ the same speed and a ratio that P4s adore.

t1mex
05-24-04, 01:00 PM
for the sake of thoroughness (is that a word?) I will post the other side:

If you are running an Athlon system, the 3-4-4-8 timings would liekly be better, because Athlons *hate* Asynchronous operation. The 5:4 would be more of a burden than a gain, even with the 2-2-2-6 timings.


Conclusion:
It depends on the system you are running, on whether or not you should go 5:4 or 1:1

Sophisticated
05-24-04, 02:14 PM
Originally posted by aNTiChRisT
Heh, its different in any PC. And specify the mhz... lol

100mhz 3-4-4-8 1:1 is worse than 200mhz 2-2-2-6 5:4 :-)

I never liked dividers, but you'll have to run some bandwidth test on it to find out.

~t0m

ummm well of course the 2-2-2-6 @ 5:4 @ 200 but thats to easy to say....how about you compare 3-4-4-8 @ 1:1 at say 230MHz then these timmings would win

mrspec3
05-24-04, 02:23 PM
http://www.pcstats.com/articleview.cfm?articleID=873

aNTiChRisT
05-24-04, 03:19 PM
Originally posted by Sophisticated


ummm well of course the 2-2-2-6 @ 5:4 @ 200 but thats to easy to say....how about you compare 3-4-4-8 @ 1:1 at say 230MHz then these timmings would win

Thats something he needed to say, rather than us speculate a load of random stats.

IMHO 200 2-2-2-5 is slower than 250 3-4-4-8, and definatley if you want to go over 250.

~t0m

NoteUser
05-24-04, 03:23 PM
I have to ask what processor and mobo do you plan to use.

If you can match the PC4000 on the same ground with 2-2-2-6 3200 the tighter timming would win.

BTW, it does sound like BH6 or BH5 but these IC's respond to voltage.

Silent Buddha
05-24-04, 06:47 PM
Originally posted by mrspec3
http://www.pcstats.com/articleview.cfm?articleID=873

Great link! It seems that running 250 FSB 1:1 @ 3-4-4-8 is about equal to running 250 FSB 5:4 with memory running 200MHz @ 2-2-2-6. Kind of interesting since I've seen BH-5 running 240Mhz @ 2-2-2-5 with 300 FSB 5:4 kill high speed memory running 300 FSB 1:1 @ 3-4-4-4-8.

Reefa_Madness
05-24-04, 09:51 PM
That PC Stats article and several others that I have read, all come pretty much to that same conclusion. Running at 5:4 with fast timings vs 1:1 at the slower timings is a wash in most cases.

What I do remember reading (it think that it was either in Tom's H/W Guide, or Anandtech) is that for applications such as video encoding, performance is better with the faster timings at 5:4 vs the 1:1 and slower timings.

This is one of those areas where everyone has a preference (and, accordingly, a bias). As has been stated before, short of benchmarking YOUR system with YOUR applications, it is one of those questions that has no clear cut answer.

Now, lets move on to an easier question...which is better, Intel or AMD?

NoteUser
05-24-04, 10:25 PM
I'm sick of seeing VIA chipsets and controllers:D How's that for an answer lol.

Intel has HT

AMD has 64

From what I hear Intel has a better chipset but AMD supposely has the better processor.

Silent Buddha
05-24-04, 10:44 PM
Originally posted by Reefa_Madness
What I do remember reading (it think that it was either in Tom's H/W Guide, or Anandtech) is that for applications such as video encoding, performance is better with the faster timings at 5:4 vs the 1:1 and slower timings.

Hmmm...that's interesting. I tend to assume that more raw memory bandwidth @ 1:1 would give better performance than 5:4 at bandwidth intensive applications like video encoding. And THG is not good for anything besides printer round-ups in my book ;) although coming from Anandtech gives it more weight.

Let's not debate AMD vs. Intel. I want to release my inner-AMD fanboyism but I'll save that for the AMD forum ;)

NoteUser
05-24-04, 11:16 PM
1:1 is faster than 5:4 naturally but it depends on the FSB you can achive and that will determine if your better off with 5:4 etc.

scanido
05-25-04, 12:05 AM
1:1 at 3-4-4-8 will 'appear' to be better in synthetic benchmarks ie: Sandra, PcMark200x, but in real world apps 2-2-2-5 at 5:4 is FASTER!

http://www.bleedinedge.com/forum/showthread.php?t=1302

The only way to beat 2-2-2-5 at 5:4 is to do 1:1 with x-2-2-x timings! OCZ's new EB line looks very promising for that.

I just got OCZ PC3700EB and will post a thread up on my findings. So far they are impressive to say the least!

Vipasnipa
05-25-04, 12:10 AM
So in the case of P4 systems, it might be better performance wise to run at 5:4 with tighter timings then it would be to run it with the higher bandwidth of 1:1 and the laxer limings?

scanido
05-25-04, 12:17 AM
Why not both?

1:1 at 250fsb 2.5-2-2-5 would be a killer setup!

From all the readings i've done, OCZ's new EB line looks amazing!

I've seen STABLE PC3700EB speeds at 250FSB x-2-2-x that will definately out pace BH5 at 5:4 timings.

Cho
05-25-04, 04:52 AM
Originally posted by scanido
Why not both?

1:1 at 250fsb 2.5-2-2-5 would be a killer setup!

From all the readings i've done, OCZ's new EB line looks amazing!

I've seen STABLE PC3700EB speeds at 250FSB x-2-2-x that will definately out pace BH5 at 5:4 timings.

1:1 at 250fsb 2.5-2-2-5 is nice for sure!
But this ram is just too expensive for me :(

Reefa_Madness
05-25-04, 06:10 AM
Yes, the EB line is looking real promising and right now is represents the best alternative to bh-5 (especially, since at the retail level, bh-5 is not generaly available anymore).

I'm not a fanboy of bh-5, or OCZ, or anybody else, I'm just a fanboy of high performance...at reasonable cost. My accounting background keeps the notion of "cost benefit" in most of my purchase decisions. Right now, I would have to say that I just can't get myself to fork over the asking price for the OCZ (or a gig of bh-5 on ebay, for that matter), no matter how good (and they are both good) the ram performs.

Everyone has their own level for "cost benefit" so my comment is not a knock on anyone else.

NoteUser
05-25-04, 01:23 PM
The prices went up on BH-6 if anyone wanted to know.

BTW, I checked the local shop site.