OCZ4000Gold IS the REAL DEAL!!!

[solofly]

There... very little difference... thank you CrocHammer...

[CrocHammer]

meowChow.

I think oldfart suggested we just start again. Clearly your ego just will not allow you to get there.

So now if someone does not agree with you they are "ignorant?" I guess that fits because you would be confiscating those computers that were not built, configured or tweaked to YOUR specifications. Of course ignorance of your hard and fast rules would make those unfortunates really deserving of confiscation. Perhaps you should post those rules throughout the forums here so people can avoid making those terrible mistakes. That would be helpful for you too as you could avoid the stress of all that railing against the rampant meme. I know people of intellect and education can be so distressed by ignorance. Good therapy to point it out where ever you cab find it.

By the way, nobody said you were wrong. Hope that helps you move on.

This is an internet forum, not a support group.

Really!! That was an unfortunate misuse of words out of context, right?

[oldfart]

276fsb 1:1 2.5,4,4,7
288fsb 5:4 2.5,3,4,7

This is one of the reasons it is so difficult to generalized about 1:1 Vs 5:4. Every config is different. For you, 1:1 or 5:4 you are still running some pretty high DDR speeds (552/460) Usually, 1:1 Vs 5:4 also means low latency Vs high latency. With your setup, you are running high latency in both setups, so that is out the window. You are looking at difference in CPU speed here.

BTW UT flyby = video card, botmatch = CPU/mem (for the most part). Use a lower game res for more of a CPU stress test, high res for video card.

Again, you have a nice setup no matter how you slice it. UT botmatch of 95+ is VERY good! Test with some more real apps (not just games) and see what works best.

[specmike]

Solofly, no hurt feelings here, I realized later who you were speaking of. I hope your question gets answered to your satisfaction.

It seems to me that at the top of EVERY page in this entire computer FORUM is the creedo: SUPPORTING COMPUTERS AND BUILDING FRIENDSHIPS Mebbe meowchow should get with the mods and CJ and tell them that they have the wrong motto on the place.

Perhaps we can change it to:
Spoonfed Computer Forum for the Thick, (Skinned and Witted.)

You could have 1 thread that tells everyone EXACTLY how to set up their rig and there would never be any need for those pesky replies, unneeded original thoughts, and annoying friendships.

And, I genuinely request a good laymans explanation (in small werds so's I can unnerstand it 2) of what memory bandwidth IS good for. I mean, why is PC3200 better than PC2100 if you have 1024mb of each with each at the same FSB? When do you not have enough or too much bandwidth? In general, we can all work under the premise that bigger or faster is better but what does bandwidth really do for us and under what circumstances do you really see it?

It's odd, I'm happy with my Corsair because of my OCing ability but what is all that bandwidth really good for?

Mike.

[P4Northwood]

Originally posted by CrocHammer
Thanks oldfart. Seems to me you are straight up too.
I am not one who builds for benchmarks but I find them useful to gauge whether the stresses of overclocking are worth the returns.

I was not ignoring you Solofly but I had to reload games to get you an answer.

Unfortunately my kid has my Q3 disc so I can't give you that. Here are my results from UT2003. I dropped both the 5:4 and 1:1 settings to ensure stability as these 1:1 and 5:4 settings are the ones I have had a chance to run more than 12 hours on Prime95 and I have not moved vcore past 1.550. It boots and operates fine at 292fsb 5:4 but I have not stressed tested at that level yet. 280fsb 1:1 runs 3dmark2001 but I am doubting the stability without more vcore. The board could really use 2.9 or 3 vdimm to help that out. I ran the 1024 x 768 so people could compare to some of the recent reviews at Anantech and the like.

276fsb 1:1 2.5,4,4,7
1024x768 flyby 244.9 botmatch 94.5
1280x960 flyby 181.3 botmatch 94.2

288fsb 5:4 2.5,3,4,7
1024x768 flyby 246.0 botmatch 96.6
1280x960 flyby 181.1 botmatch 95.2

I don't know what to say. It seems that at the 1280x960 there is little real difference. I am currently gaming at 1280 x 1024 with all eye candy on. The ATI9800Pro is new to me so I have not done any overclocking or playing with the enhanced settings yet.

Maybe I could try some 1600x1200 to see what happens there.

I don't know what to expect out of the core and mem on the card. I think from the Samsung id's on the chips they are 3 nanosecond and not the ultimate 2.8 nanosecond chips.

Bottom line it is a pretty fast rig by more luck than any major strategy on my part. It's a tough call for people as far as ram is concerned. If I actually knew in advance I was limited to say 275fsb 5:4 I would probably go for that at lower timings if I could reliably find some memory that would run at the 224 to 233mhz level.

Crochammer, I have the same issue as far as 292FSB@5:4. I wonder if this is all board related. I wonder if the P4C800-E Deluxe has an hidden issue at around that FSB. Some of your argument makes alot of since to me. I am almost having the same issue. Our setup is almost the same, although you have 512mb more memory than I do. However, it is with 2 sticks as well. So I am wondering if the Asus board would like all four slots populated to gain extra stability over 292FSB. I know both of our memory cant be stumped at the same FSB. I doubt very seriously that it is the voltage of the dimms, because even on my MSI board I did not witness any advantage to the increased voltage. On the contrary, it seemed to favor the 2.85v.

So do you think there could possibly be a flaw of the P4C800-E above 292fsb. Or is there more than 1 documented case of clocks well over 300fsb. Thanks Croc for you research effort.

[CrocHammer]

OK, peace brothers. meowChow caught me with a bad headache. I usually let that stuff fly by and just wave at it. Thanks for the reminder Mike.

oldfart.

Hindsight being 20/20, if I had known I would happen upon this 2.4C I might have jumped on some memory with bh-5 chips. At the time I could not see much available in double-sided anything that was price worthy. OCZ did well by me and sent out the 2 x 512's so how could I not be happy.

I may very well be able to drop the timings on the 4000Gold at 5:4 timings but as you can see by the results, I did not see much point. There is a lot to be said for FAST and STABLE. Yes, I could try dropping the resolution to stress test both the vid card and the cpu but I just don't see the point of that for me except for the information.

My intent was to let people know how the 4000Gold did in the real world given I did not have engineering samples, had a "luck o' the draw processor and an off the shelf mobo.

Now I would pretty much just like to enjoy my good fortune.

[oldfart]

This is why I went with a 2.6C instead of a 2.4C. The 2.4C gets sky high in FSB speed (if you are lucky and get a good one!). The 2.6C will have a more reasonable FSB @ the same CPU speed (again, if you get a good one)

I think 4 sticks would be worse for stability, and you lose PAT.

If I were you, I'd just "suffer" with the 3.5 GHz that you have.

[CrocHammer]

I think Prometheus ( Wesley Fink ) over at Anandtech found a glitch in the board that seemed to shutdown PAM when all 4 slots were populated with doubled sided dimms but otherwise try his article, it's a good one.

Here

I also very much enjoyed Wesley's review of intermediate memory which really rounds out the debate. Certainly, this review picks up on meowChow's take on things.

Here

Hope all that helps out. There is certainly more to choose from now than a month ago.

[CrocHammer]

Thanks OldFart.

Yes I have been chewing on that dilemma.

I still have the 2.6C that will struggle to 258fsb 5:4 and 250 1:1 with the same dimms. I will probably keep the 2.4C and pocket the $120 difference.

[oldfart]

Your 2.6C sounds like mine. 257 FSB is where it is happy. It may go a bit higher, but it is in the "lots more Vcore for a few more MHz" curve. I like to keep the Vcore reasonable.

[CrocHammer]

Yes exactly. I really needed 1.625 vcore for the 258fsb.

With the 2.4C it shakes it's head at anything above default vcore until I get to 288. It is also running cooler than the 2.6C which I am always happy to see.

By the way, this is not intended as a gratuitous plug but OCZ sent me some of their OCZUltra ll silver heat sink paste. I am using that right now. It is doing a few C better than Ceramique already.

[oldfart]

That 2.4C sounds like a keeper to me.

[CrocHammer]

OK I got my Quake 3 back and loaded it up and patched to 1.31.
I used the standard DM4 demo at 1024x768 and 1280X1024 and had highest settings on including the game section for all eye candy at both resolutions.

1024x768
276 1:1 416
288 5:4 399

1280x1024
276 1:1 384
288 5:4 382

There were some wierd numbers at 276 so the 416 is best of 3 runs. I never paid much attention to Q3 before for benchmarks. I had to do a Google search to find out how to run the demo.

Anything else you would like to see that will help?

[solofly]

Because I have a 2.4 chip, FSB gets out of the range for most memory sticks. If I get some PC4000 stuff and end up at 280mhz, I know I won't be happy to downgrade from 290FSB where I sit as we speak. Now if I had a 3-gig chip that would be a different story and 1:1 ratio would be more than likely. So, I need a set that will do 232mhz (464) with timing of at least 2-3-3-7 at 2.75v or 2.85v max. OCZ carries two sets that fit this category. The ultimate set would be 2 x 512 OCZ PC3700 EL Gold but they cost more than PC4000. The other one is OCZ PC3500 EL (standard) and then I would overclock it. Most reviews I read reach 245+ (490) with these sticks. Amazing for a lot of less cash. Good idea or bad idea?

[CrocHammer]

Solofly.

That's a great snapshot of the decision a lot of people are struggling with.

For me it was easy. I was was not going to spend the $'s on a 3c processor. My history has been 700E at 1 gig, then, 1.8A at 2.5 gigs, 2.6C at 3.3 gigs and now 2.4C at 3.5 gigs. As much as meowchow does not understand it, there is a kick to getting that out of a 2.4C processor that some folks enjoy just for the sake of it.

As much as the debate about 3c at 1:1 rather than 2.4c at 5:4 at 3.4 gigs is interesting and for some a matter of evangelical dogma, the real truth is, either will get yah there for the most part. Few people talk about whether your inherent hand-eye co-ordination gifts gets you more Q3 kills in the "realworld".

I don't think OCZ has 512meg sticks in the 3700Gold. Maybe they do now but I would check first. If I was going 2.4c I would go with the 4000Gold or maybe even the Copper. The 4000Gold is not far off the timings if you went 5:4 to get there. They are good to go for me at 2.5,3,4,7 and I have not pushed harder yet.

The 3500El is a good idea only if it works out and there are no guarantees. Regrets are not for overclockers, right?

If you went 3C the new Muskin 3500 stuff is using the Winbond bh-5 chips. Tight timings for the minimal overclocking you would expect. Prometheues ( Wesley Fink ) just did a review with that and others over at Anandtech and he IS an expert. Check it out.

Good luck solofly.