The OCForums 2x1Gb v/s 512x2Mb DDR1 Bandwidth challenge!

[Super Nade]

DDR1 ONLY! No DDR2!
​

Folks,

I believe it is the right time to offer a challenge to the 2 x 1Gb camp. Since a lot of people are opting for 2Gb kits, let us see if 2Gb kits deliver the killer punch they promise over 1Gb sets.

I've seen many threads in these and other forums claiming that 2 x 1Gb offers better bandwidth than 512 x 2 Mb. Some people have claimed that their 2 x 1Gb sets do better than both TCCD and BH-5 maxed out. I'd like to know how true these claims are. Since I don't own a high performance 2 x 1Gb set, I'll have tor rely on those who do, for results. There is only one way to find out isn't it? Given below, are a few simple rules to be followed. So, go on, whip out your best kits and push them to the max. Everybody is welcome to participate (both AMD and Intel).



Rules:

• You MUST post SPi 32M (use the millisecond modded version and make sure the checksum is visible) screenshots as proof of stability. Nothing less will be accepted.

• CPUz and A64 Tweaker screens are a must.

• Synthetic benchmarking suites to be used are ScienceMark, Everest and 3DM01

• Real world Benchmarks may include Winrar or anything else of your choice. You may post game benchmarks or ss's showing fps, but that is not necessary as not everybody owns the same games.

I realize that certain benchmarks do depend on the CPU, so if you have any better ones, please feel free to share the info with me. I'll add them to the list as and when necessary.

Let the games begin!




Top Bandwidth scores { ScienceMark/Everest/Sandra} : ( 2 x 1 Gb, 512 X 2 Mb)

1. RedDragonXXX, 300MHz 3-3-8-1T (Crucial Ballistix PC4000)

2. Super Nade 270MHz 2-2-2-2, 1T (Mushkin Lvl 2 Blk PC3500)

3. Deathman 285MHz 3-3-3-8-1T (Crucial Ballistix PC4000)





















--------------------------------------------------------------------------

 

[(cf)eclipse]

2x512mb will score higher simply because sPI doesn't use enough memory.. hell, if you had an option for 2x256mb, it'd probably take the honors. the difference isn't the bandwidth and latency as much as the reduction of hard drive accesses with 2gb, which is significantly slower than getting something out of the ram


that said, let the games begin

 

[Super Nade]

If you look closely, I'm suing SPi to ensure that there are no kamikaze benchmarks . It is just for proof of Stability

The reason I threw this challenge was because of this article in techpowerup:
http://www.techpowerup.com/reviews/Mushkin/XP4000-2GB/2

Now this memory is very versatile, its got some really great bandwidth and timings for a 2 GB kit. At 230 MHz it was running timings tighter than most TCCD and TCC5. At 265 is was running with timings tighter than some BH-5 and all TCCD and TCC5 chips. I found the RAM to be max. Video Bench stable at 280 and SuperPi and Bandwith benchmark stable at 286. My results were done on a i865 platform which doesn't have a strong memory controller like the Athlon64, FX, and X2 series of AMD processors. But just these results prove that this memory is great on all platforms.

Let us look at the colored sentances.

1.Who would want to run BH-5 and TCCD at 230Mhz? What did the author mean by saying those modules ran at tighter timings than BH-5 @230? I will prove (shortly) that my BH-5 will run at 1.5-2-2-2-1T@ 230Mhz. Can those modules do that? I'm not so sure.

2.At 265MHz, from the table published in the article those 2Gb modules ran at 2.5-3-2-5 2.8V. Do you call those timings tight leave alone tighter than BH/TCCD?

3.How can you "prove" that those results are great on all platforms if you hvae tested it only on one platform (by platforms I mean AMD v/s Intel)? I would tend to agree if the author concluded with an expectation of better performance, but I don't think one can generalize without an actual test. The author concludes by saying these are awesome for overclocking. Based on those results, I fail to understand what is so great about them?

I have seen similar claims flying about regarding XXX RAM being the best ever in the whole wide universe........

Hence the open challenge. If 2Gb modules offer better bandwidth, that would be awesome! If not, people would know that things are not yet up there yet, to warrant a big hue and cry.

 

[ochungry]

That article is going to create liability for Mushkin, and flood of RMA's, thereafter. When the claim is not achieved by end users, specially the AMD enthusiasts, it is then Mushkin will realize the fiasco that article has caused.

 

[Reefa_Madness]

Well, this should be interesting and informative. I'm sure that many will be following this thread with great interest.

 

[cullam3n]

Hmm... sounds like a plan for tomorrow afternoon

 

[Super Nade]

That article is going to create liability for Mushkin, and flood of RMA's, thereafter. When the claim is not achieved by end users, specially the AMD enthusiasts, it is then Mushkin will realize the fiasco that article has caused.

Well, let us not jump the gun yet.
This is an open challenge to all companies marketing 2Gb sets, not just Mushkin alone.

 

[Son1990]

Looks like now I have a reason to buy a 2gb set. Mushkin Bluelines, here I come.

 

[deathman20]

Oh cool I'll have to do this later today. I got basically all that info just no PI doing 32MB run.

 

[ochungry]

Well, let us not jump the gun yet.
This is an open challenge to all companies marketing 2Gb sets, not just Mushkin alone.

No I am not jumping the gun. I am saying every time some review comes out and people don’t get the same results, it causes a lot of headaches for the reviewer and the company being reviewed. The reviewer may be better skilled than most end users and the equipment used could also be determining factor. I have read Steven's reviews on other items on that website, and it appears that he is very knowledgeable and skilled in whatever reviewing(or modding).
This is the problem. Most of us are not as skilled, and thus causing the commotion and sometimes bad observations. your challenge here, could have stemmed from your blind devotion toward TCCD. So I do not think you could be as objective either when criticizing someone who challenges your review or belief. This is another problem that I rather leave alone.

 

[cullam3n]

Can you post a link to this millisecond SuperPI?

 

[Super Nade]

Cullam3n here is the link:
http://superpi.radeonx.com/



ochungry:
Could you please elaborate on that? How did you conclude I was blindly devoted to TCCD. As of today, I do not own a single set of TCCD. All my sets are vintage BH-5 and one set of UTT (which is garbage).

Now that you question my objectivity, I would like to know on what factors do you base this statement? XYZ claims that 2Gb modules offers better bandwidth than TCCD or BH-5 at the same frequency. I say prove it. Simple as that. Where is my objectivity come in here? If you believe that the reviewers are super special and extra skilled, then you are wrong. The overclocking community does not lack skill. There are numerous people as skilled as website based reviewers. To put it bluntly, it takes less skill and more patience. If you have the patience, you can get the best out of your system. The only skill which seperates the gurus from the average enthusiast is the analysis and interpretation. O.K, you have ABC for your results, you could either state them blandly or offer valuable insight via the mechanics of the process. Less that 1% of the reviews I have read offer such insight.

So, before you question my objectivity, try to frame a coherent argument to support your claim.

 

[ochungry]

Super Nade your colored quote mentions TCCD/TCCX. and I have seen how you defend TCCD in numerous occasions. You creating a thread challenging a review on another ram that has compared it to TCCD brought me to this reply.
I use TCCD myself and used to be a fanatic, but lately I have seen there are other options too. Besides, that review was on 2x1gig, I didn’t think TCCD's come in 1 gig anyway.
If it is BH-5 that incited you to this challenge, then why mention TCCD that aren’t made in 2x1gig.
As said I rather leave it alone.
I would have liked your post not having sounded so much like a duel between a particular brand versus another particular brand(s), mainly TCCD, BH-5, and Mushkin. This is where objectivity comes to play.

Originally Posted by The Reviewer
Now this memory is very versatile, its got some really great bandwidth and timings for a 2 GB kit. At 230 MHz it was running timings tighter than most TCCD and TCC5. At 265 is was running with timings tighter than some BH-5 and all TCCD and TCC5 chips. I found the RAM to be max. Video Bench stable at 280 and SuperPi and Bandwith benchmark stable at 286. My results were done on a i865 platform which doesn't have a strong memory controller like the Athlon64, FX, and X2 series of AMD processors. But just these results prove that this memory is great on all platforms.

Some people have claimed that their 2 x 1Gb sets do better than both TCCD and BH-5 maxed out. I'd like to know how true these claims are. Since I don't own a high performance 2 x 1Gb set, I'll have tor rely on those who do, for results

 

[Super Nade]

Mate,

Everybody knows that TCCD and BH-x do not come as 1Gb sets. I'm I'm not mistaken, I heard that GSkill were attempting to put out a TCCx (x=! 5 or D) 2 x 1Gb set.

Well, to clear it up let me state this. I have not seen any IC garner higher bandwidth than TCCD or BH-5, hence I am skeptical of any claim made to the contrary. Does this mean I'm a TCCD/BH-5 fanatic? Not really, I'm just expressing surprise about the claims that high density modules can match a)Insane clocks of TCCx at CAS 2.5-3-x-x-1T or b)The insanse 2-2-2-x-1T timings of BH-5 at say 270Mhz.

If I see proof, hey I'll be very tempted to go buy my next set of 2 x 1Gb just for the heck of it!

I still don't see the connection to my objectivity? You call somebody's objectivity into question if they have an stake in the experiment. What stake do I have? I don't work for Mushkin or anybody else?

All I'm doing is trying to find out if what the reviewer has said was a general trend (if it is I need to keep myself updated!)?

I don't understand your questions mate. Somebody claims 2 x 1Gb offers better bandwidth than TCCD or BH-5 maxed out, I said show me the proof. Claims demand answers to back them up. Its just that simple. Where is the conudrum here? I'm afraid I can't make myself anymore clear than this.

If I offended you with my reply, I apologise. That was not my intention.



S-N

 

[ochungry]

Super Nade, you did not offend me. And I did not mean to sound offensive either. I said I do not think you "could" be objective, which is not concrete statement, but only implying. I apologies if sounded doubtful.
This thread will help choosing the right memory modules for different needs and budgets. Thanks for the effort.

 

[deathman20]

Was going to post last night then realized that I forgot to get the millisecond version after i ran the 32MB test So I played some games afterwords instead of redoing the test.

Anyways I guess i'll try it again tonight.

Now the tests I plan to have on the screen are:
Sandra (MIPS/Media/Ram)
CPUZ (4 windows showing the 2 cpus/and the 2 mem screens)
PI with results
3DMark05 w/ my oc on the card+voltages shown

Is that fine to be added for this?

 

[RedDragonXXX]

2x1GB Crucial Balistix Micron 4Y/5B D's

 

[Super Nade]

Looks good.
Mate can you post a few more screens i.e Sciencemark and Everest Read,Write and Latency? Thanks!

 

[RedDragonXXX]

Don't have a vid card and I'm waiting for the RAM to return from RMA because one of the sticks was deffective.

 

[Super Nade]

Was going to post last night then realized that I forgot to get the millisecond version after i ran the 32MB test So I played some games afterwords instead of redoing the test.

Anyways I guess i'll try it again tonight.

Now the tests I plan to have on the screen are:
Sandra (MIPS/Media/Ram)
CPUZ (4 windows showing the 2 cpus/and the 2 mem screens)
PI with results
3DMark05 w/ my oc on the card+voltages shown

Is that fine to be added for this?

No problem man. If you don't have the millisecond version, just post the one you have. Also try to include Everest and Sciencemark scores. I like Sciencemark for the rather detailed scores one can see and everest is what I've been using for a long time.