Printable View
I did some testing and so far the results are a bit disappointing...
My previous PC3200 VX did well on HTT 262 with 2-2-2-3 timings.
But the VX PC4000 does only HTT 253 wih 2-2-2-5 timings.
Kind of frustrated...
Wish I havn't sold my VX PC3200 a few days ago... :(
People who got there PC4000 VX please post your results...
have you burned in your new VX and was that max after you burned in your old VX?
Yup, I've burned-in for guite a long time.
Just as I did with my old VX. :(
Quote:
Originally Posted by nebuchanezzar
How much voltage did you need with your PC3200, and how much with this new PC4000?
My old VX required 3.6V for HTT 262, but my new VX needs 3.4V for HTT 253.
Though as guaranteed HTT 250 with 3.3V works well.
The only problem is that with whatever voltage, I can't get over HTT 253. :(
I did the testing with my NF4 SLI-DR mobo,
which I used before with my old VX.
Quote:
Originally Posted by mcnbns
Have you played with different tras, trc, trfc and tref settings on this new stuff?
Sorry to hear that, man. Maybe you just got a bad pair? OCZ might let you RMA it, because those are some really crappy oc'ing results... 3MHz over stock with +0.3V? :/
Yup, I tried various settings for my new VX.
And so far nothing helps. :(
Some of my friends here also got there new VXs,
so I'll be getting more results to compare with.
Will post the results here. :(
deam, only 250mhz?
so its decidedly bad deal :mad:
my mx can do 270 2-2-2-5 stable and they 3200.
I've got 3 more test results from my friends who bought there VX PC4000.
They too, are only able to push the ram around 250-255.
(Now I'm starting to think that my new VX is not bad at all :( ,
just an average pair)
I don't think HTT 250~255 with 2-2-2-5 timing is bad at all,
compared to the old BH5 stuff.
But I had high expectations.
I mean people saying that the new stuff would be speed bined
somewhere above the 250 point...
Guess I had too much expectations with the new VX.
250~255 with 2-2-2-5 timing is still nice compared to a lot of rams
out in the market, but I feel a bit bitter for OCZ releasing
a product that shows not much of a difference with only
the product name changed.
Guess the people who has nice PC3200 VX that can do over 260
should not sell them and stick with there old PC3200 VX.
Man this is an outrage! Anyone tested the "new VX3200"? Or value VX? to see if they oc even "poorer" (jeez I sound like a spoiled kid, but whatnot). Maybe (hopefully - even though it sounds even meaner :D) just bad week?
Edit: Somehow I feel I didn't put that very well, but hey, you get what I mean... :stick:
At least you can get to 250. Mine keeps getting errors in mestest(test 7) and they just won't go away. I have to run it at 245. :(
Hey, if you can't get 250 (DDR500) stable you can RMA it. You sure you're speaking of VX PC4000 and not that VX PC3200?Quote:
Originally Posted by GregP24
oh man....
i've got twinmos SP with UTT .... sucker.... they need 3.2 v for 243 mhz completely memtest stable BUT!!!! I CAN'T RAISE THE FSB HIGHER THAN 244 at all.... 3.6v doesnt help.... i've also testet 3.9 volt.... i am going crazy with this stick.... can you guys help me? I've tried to burn in but it doesn't work for me....
I've got a MSI k8n neo2 plat. and 3200+ winchester... and ddr booster I know, my board and CPU can reach higher fsb, because i've tested it up to 310 with the tccd's of a friend of mine...
help
thx
Rondrian -
The Neo2 can be a funny animal at times, ramp the volts to 3.5-3.6 and try a suicide clockgen up to 260.....worked for me and my BH5 - I had instant prime falier at anything above 235-240, but solid as a rock at 260 (3.5v).
Edit: Typo
Sorry, guess I should have specified. PC3200. It's annoying, though. I see all these people hitting 260-265@2-2-2-6 on the PC3200 and I can't even get close. My TCCD Plat Rev2 was faster because it would run 270@2.5-3-3-7.Quote:
Originally Posted by krille
Hey guys, OCZ does their best, If you bought PC3200 VX that does over 250 mhz, don't sell that and buy PC4000 VX. We test for some headroom at rated speeds, but our rigs are different than yours. If you bought PC4000, expect 250 mhz at rated timings and hopefully a bit more. We can't force the chips to go faster (maybe you can) and we test for rated speeds. Buying PC4000 gives you some insurance about speeds. If we could get PC4400 speeds out of these we would be selling that.
Just the facts of life.
That said we appreciate your business and want you all to be happy.
Remember, the only reason it seems ocz released these, is because people were getting around 250 with 3200's, so now basically theyll be binning for 250 and not caring about above, appart from getting cherrys, so as with all higher rated stuff its not going to o/c as much, all it really does is ensures you can get to 250. But as i say, its basically the same product.
I'm not trying to complain. I'm happy with the mem. I just wish I'd gotten some stick that will go to 265. We all want the uber OCing equipment and are always dissapointed when we don't get it. I do love OCZ products!
hey Andy, do you binn them at exactly 250? don't you usually binn the stuff a little above specs?
thx
Quote:
Originally Posted by krille
They do, but you have to read between the lines. Their testing platforms are different from yours, so while they will garuntee 250, they can't really garuntee more based on the differing test platforms.
Generally there is headroom tested in. Like I said our A64 memroy controllers might be better or worse than yours. Throw individual boards and possible power supplies into the mix and who knows what will happen.Quote:
Originally Posted by krille
I would say VX at 260-265 would put up a very strong showing against TCCD at 300-305. Its just a matter or latencies.
JazzMang Out.
Yea doesnt seem like OCZ binned this batch properly :stick:Quote:
Originally Posted by cupra
I hope the next batch will turn out better
doesn't work for me :(Quote:
Originally Posted by Theo404
One person not happy and we haven't binned the whole batch correctly :rolleyes: :eek: some how i can't see this being the case.Quote:
Originally Posted by Sentential
Many things dictate how well ram overclocks on A64, one of the main issues is people expecting way to much from winchesters, fact is the cores do good clocks, the memory controllers don't.
Set a much lower multi when your testing, IE 6 or 7 and see how far the ram clocks 100% stable, then push the multi up till the memory controller fails, thats the best way of finding where you will end up running your rig.
Sorry to hear that Rondrian.
8 sticks may not be a batch, but come on, only 5 or so mhz over stock for 4 different cases....seems a little weak. Do you think RMA will be an option for these unlucky folk? I know OCZ as having the best RMA dept in the buisness (have read threads at bleeding edge about people being cross-shipped much better ram then they bought years after purchase) so I guess its a possibility.
AFAIK they garuantee you ddr500 (if you ememory controller can hadle it) anything beyond that is a bonus why should they be liable if you cant hit much higher. Considering there are so many other variables in play, your cpu, mobo, psu etc.
The biggest variable here is the memory controller. I had a Neo2 and a Winchester that was horrible. The memory controller did about 255mhz with that board. I put that very same CPU into the DFI NF4 Ultra-D and it does 300mhz with TCCD. It will run 260mhz with VX with no problem, but I REALLY had to tweak the crap out of it to get it there. The DFI is probably one of the toughest boards I have ever used. Every CPU you put in seems to need different timings. The timings you see posted around should be a starting point only and not gospel. OPP's timing more than likely will not work for you unless you have the vrtual twin to his FX55 in your DFI.
Spend hours on tweaking, write down what you have tried. Try different combinations, recording what works in memtest without errors. You will get there eventually. Don't forget to burn the VX in for 48 hours or so at 2-2-2, with 3.1v-3.2v at 200mhz.
You will find many many DFI users with RAM from every company in the business blaming the memory for their errors and issues. Even guys that had the memrory running great in their old setup blame the RAM, because it suddenly must be bad.
I could go on about at having at least a 500w PSU with a minimum of 28A on the 12v rail also. The possibilties for errors don't start and end with memory.
RMA's cost you money. They cost the manufacturer even more money. We are trying to keep the guaranteed speeds up and the prices down, so please be patient with your memory, no matter what brand.
With the new speed binning of PC4000 VX does that mean the regular PC3200 VX wont do clocks above 250 unless it came out before the new PC4000 release?
they might just take more volts
hmm wondering how this plays out
wow, so many disgruntled customers...
first of all, let me start off saying i was never an ocz fan so im not defending ocz in any way, i was always a mushkin fan because of the level II black pc3500 bh-5...it was these sticks that won me over back in the day
it was only recently that i decided mushkin sucked because they charged so much for their tccd vs other competition tccd, which fared better in many reviews over the mushkin anyways
so i made the jump from mushkin to ocz and here i am
that being said, we as overclockers should know that an overclock is NEVER a guarantee. ocz guarantees a certain speed and binned it as so at a specific voltage...they do NOT guarantee an overclock or even a guaranteed certain percentage above stock speeds! or they would sell it at that speed as someone said earlier
an example would be an A64 4000+ that will maybe hit around the same ceiling as some good winchesters 3000+ or 3200+'s
just cuz the 800 dollar chip hits the same ceiling as the 150 dollar chip doesnt merit the right to rma the 800 dollar chip because you were expecting 3ghz
btw, i duno how much the 4000+ costs, just throwing out a figure
No internet connection for a few days, and here I see lots of posts.
I'm not sure on what to say first, but back for a few days,
I'v done lots of tests with 4 pairs of VX PC4000.
I borrowed 3 pairs from my friends and tested them all, because I was
the only person able to run 262 HTT stable at 2-2-2-5 with my current rig.
Same mobo, cpu, psu as before with my old VX PC3200,
and the outcome is pretty much the same.
The 4 pairs I have hit around 252-255.
One funny thing is that out of 8 sticks,
2 sticks don't even like voltages. These 2 will only take voltages upto 3.4V.
Kind of awkward since the product's name is Voltage Extream...
(Well, 3.4V is still lots of voltage compared to ordinary ram)
I'm not trying to prove somthing like the new VX PC4000
aren't good or something.
I mean I still think HTT 252-255 with 2-2-2-5 is great,
and only OCZ is able to make these kind of great products.
One single problem is that my expectations were a bit too high.
Because I've had 3 pairs of VX PC3200 and all of them hit over HTT 260. :)
And also I'm not the only one with VX PC4000s.
Maybe I was just a bit unlucky. I've had 3 pairs of VX PC3200 doing over
HTT 260, and that probably made all of my luck go away. :D
Well, recently I'v RMAed 2 pairs of EL PC3200 Rev.2 TCCD,
and had the greatest RMA experience.
And one pair I got back is able to do HTT 295 2.5-3-3-5. FANTASTIC!
Gotta tell ya I'm still a fan of OCZ.
Let's hope that OCZ will make even better products next time. :)
OK guys, please take the time to read this:
We need ALL of you to understand some things.
1) You need to "burn in" your VX memory in for about 48 hours at 200mhz with 3.2v and 2-2-2 timings. Then increase voltage and speeds over a few days. This helps greatly. Then, even over time the voltage required seems to drop about .1v and the speeds will improve.
2) We test thoroughly and allow for some overclocking headroom. However, we don't have your CPU, motherboard, PSU and overclockling skills to verify the RAM with before your purchase. If we sent the same kit of VX to 10 guys, i'll bet 7 would be happy and 3 would say the memory is no good.
3) Some A64 CPU's have poor memory controllers. Some Winchester cores are the worst. Remember than an A64 CPU has the memory controller on die. There is no chipset memory controller like Intels or NF2. Your CPU might overclock great, yet may have a horrible memory controller. If the memory controller can 't run 2-2-2 at 260+ you will never get there. 2.5-3-3 at 275mhz is not the same as 2-2-2 at 260mhz. Tighter timings put more stress on the memory controller. I ran the same 3200+ Winchester CPU in several different boards and went from crap to good and back again with no apparent reason.
4) DO NOT use the 5v jumper on the DFI NF4 board with a 20 pin PSU. This will not work properly, DFI does not recomend it and you will never get rid of the errors if you can boot.
5) Cool you RAM. Running all this voltage is great, but witout a strong fan blowing directly on the memory there is nothing to carry the heat away. You will have errors with no or improper cooling, guaranteed.
6) DFI NF4 boards using the 5v memory jumper must have the MOSFET right near the memory in the corner of the board, wearing that big aluminum heatsink cooled. The thing gets HOT. Position a fan to blow on it also. This can cause memory errors as well as other haet related problems.
The only way you can be certain of getting anywhere near a stable overclock above 250mhz with ANY memeory is to have your RAM cooled and your timings correct (This is absolutely crtical on the DFI NF4. You CANNOT plug and play with this board. If you want to plug and play and can't be bothered with timings, SELL ITand buy something easy, like a MAC Mini :))
After you have considered all of this, if you are sure an RMA will get you from 252-255mhz up to 260mhz, email me with your complete system specs and if you don't have an glaring problems and are set up properly, I will figure out a solution for you. Again there are no gurantees that a retested set that does 260mhz on our test rig will hit 260mhz on your board. We set up our rigs with off the shelf motherbaords and lower rated CPU's to simulate the average users system.
The bottom line here is that we want every customer to be happy. We do what we can to insure that. We thought that guranteeing you 250mhz would be a good thing. That way you won't get PC3200 that "only" hits 245mhz and be upset. The extra testing add costs, and we have even tried to work the prices down for you.
Thanks for you time! We do out best for you.
Quote:
Originally Posted by andyOCZ
i duno if thats a good idea...lol..quick edit it out!
edit: i guess im an ocz fanboy now with support like that :toast:
It was no mistake, but read the entire post carefully. B4 putting yourself and OCZ through this trouble an expense, be sure of your sytem, the way you built it and it's capablities.
try playing with the drive strength options in the bios. These settings are critical and can make it or break a ram overclock on the DFI board. Also increase your chipset and vcore a bit. like andy said, make a data sheet and go into memtest86 and keep running test5 until you get errors. Change a drive strength setting and write down what you changed and run the test again. People expect to just pop VX or TCCD in and get amazing clocks. I however am not buying the memcontroller excuse since you were running your pc3200 at higher speeds. I really think it's a matter of burning in and of course you should always have a fan over the dimms. You need to be patient. And then there's the pc4000 question. Overclocking to me is taking the cheap :banana::banana::banana::banana::banana: and making it run as fast or faster than Mr. Moneybags on Park Avenue with the FX and refurbished Delorean. You should've kept your pc3200 and bought your girl some new shoes. :rolleyes:
OCZ's guy have 100% reason....i got 3200VX and this is my result
3500+ Winchester....MSI Neo 2....max stable are around 255 2/2/2
3500+ Newcastle...same Neo 2....max stable 265 2/2/2
continue your good work OCZ :D
.Quote:
Originally Posted by Boogotop
Amen! OCZ is always making sacrifices to the overclock gods so we don't have to.:toast:
How can anyone complain about a product that meets standard?
If they called it VX5200, then you've got a gripe....
Here's the bottom line:
these UTT chips work really well, throw them on just about any PCB and they'll do well. Probably 50% of all VX3200 would meet VX4000 standard.
Probably 50% of all VX4000 would pass as VX4200....
When we get a bad CPU, we sell it and buy a new one.....
Mem sticks are the same.... and for that matter, they may work well if you burn them in correctly.
C
Agreed, I hate it when people complain about only :rolleyes: getting such and such overclock. There are a million people out there that complain every day that they can only reach 2400 on their 3000+ winchester. I don't think they realize that they just got 600 MHz free. Overclocking is not guaranteed, and I wish people would realize this. If you have the money, just keep buying and selling your parts until you get a part that overclocks well. Otherwise, be happy it runs at spec!
There are good reasons to RMA your RAM if you ABSOLUTELY can not get it to run at spec.
For example, I bought OCZ 3700EL Platinum TCCD that was rated for 233 @ 2-3-3-8 with 2.8V. No matter what I did, I could not get it to run at that. I talked with an OCZ rep and and tried to figure out what could be the problem. At the end of the day, we still couldn't get it to work, so it was RMAed.
Not meeting spec = relatively good reason to RMA.
Not being able to get past spec = poor reason to RMA.
On a side note, someone mentioned that some people have RMAed to get back better ram. Somehow I RMAed and got back worse ram... 3700 Gold Rev.3. :( Seems an error occured and I got shipped the wrong ram, but again OCZ came through and is sending another replacement. :toast:
Can't you see the point of this guys? He had 3200 @ 262, now on the same baord/chip the 4000 runs slower. I think we'd all be a little dissapointed.
It's good that it runs at spec but what are we all doing here? Not picking components that run stock speeds thats for sure. I think he is doing us a service as well by letting us know his results.
Still kudo's to OCZ for staying on top of this and offering the kind of support that they do:D
Yah I understand what you're saying, but it's luck of the draw sometimes. Take for example video cards: I can overclock my 6200 to past 6600 GT stock core speeds, but some people with 6600 GT's can't even reach my core speeds overclocked! When you play the overclocking game, you gotta expect that no matter what you buy it may not overclock well. Unless you buy a CPU/memory/video card that is rated PAST what you previously could overclock, don't sell it and buy another one that is rated for lower than that! There will be a chance that you can not exceed the previous overclock you got, that's just a fact of life. A cheaper CPU/memory/video card may clock better than what you have. My 3200+ only gets to 2350, but there are a million 3000+ that do 2400+. If I have a 3000+ that does 2600 then sell it and buy a 3200+ expecting to get over 2600 then get upset when it doesn't, in this case I can understand the frustration, but it definitely goes with the territory and should be expected. Overclocking is a game.Quote:
Originally Posted by G H Z
It's luck of the draw guys, luck of the draw...
My major concern is that he and 3 of his friends all got "rated-only" sticks. What are the odds of that happened? Unless, the RAMs are coincidentally all coming from the same bad batch? Or, they really are speed binned very, very tight? Either case, it does not look good to me.
mine arrived today ;)
can't test them atm as mobo is away for voltmodding (after I killed the last one I decided to have it done prof. this time :D)
Guys most of the issues im seeing here is because we released the VX at PC4000, you atomatically expected it to clock higher than the 3200 did. We NEVER guaranteed the 3200 would hit ddr500+, infact i leaked out the test spec and got a stiff telling off for doing so. We have tried to give you a guaranteed ddr500 at 2-2-2 with the 4000, if it doesn't meet this spec and you have taken all of Andy's advice on board and it still hasn't helped then OCZ will look after you...thats what we are here for ;)
If you feel launching VX at 4000 speed was a bad move, then please try our 3200, it still clocks well and with burn in may give the same speeds or close to it, remember though its not guaranteed to do 500 so it would be hard to help you if your at 248max or so ;)
My personal old VX samples, do 260 with my 3800+ newcastle, 264 with my Winchester and 268 with my 4000+, they are well burn't in and maxed out now. You see an 8fsb difference there, thats same board, same ram, same psu...just different cpu...so the cpu has a lot of influence, you have to hit the happy medium with what the memory controller will do.
Now a few have told me increasing TRAS to 7 or so will gain you a few extra fsb and also TRC at 9 and TRFC at 24 can help also, this is not an exact science though, you will have to test and compare on your own rigs.
when i want to burn them in: has it to be done at once or is 4*12h as good as 1*48h?
just got my vx pc4000
everything above 220Mhz @ 3.2V gives my plenty of errors :confused:
A64 3500+
DFI NF4 SLI Dr
1GB Ocz vX
Most of us have noticed that it takes more than a total of 48 hours to burn VX modules in. You can do it gradually over time. What I used to do was just run Memtest Test #5 overnight daily for several weeks. There isn't any definitive source that discloses the length of time required for "burn-in." Just keep pushing until you hit a hall. ;)Quote:
Originally Posted by xsky
How long have you been using the VX? Do you have active cooling for the RAM? Try testing each module individually to rule out the possibility of one stick being defective.Quote:
Originally Posted by joe smith
To start with don't use 3.2v at 220mhz. They are not rated for that. Use 3.4-3.5v and set you timings up correctly. Just setting to AUTO will not work. Use 209 or 218 BIOS and see the Xtreme Bandwith Forum DFI Timings sticky for some suggestions.Quote:
Originally Posted by joe smith
Use fans on the memory and on the MOSFET with the aluminum HS by the RAM. Heat will produce errors and failures.
Keep tweaking, you will get there. :)
Yeah i'll try that. I thought they would do more than 220 with 3.2V.
I'm using the settings posted by bigtoe on the ocz forums.
Right now i'm using a big extra fan (~20cm, not a pc fan) which is pushing a LOT of air on the dimms and i'm having PWM temps of 32°C and CPU @ 27°C (smartguardian, full load)
I also tried a lower multi (10x, 9x) but at 221 i do get errors
My burn-in procedure isn't completed yet. Maybe i need some more hours, but i thought the mhz gain would be less than 15Mhz.
My bios settings :
Dram Freq. Set : 200
Command per Clocl : Auto
TCL : 2.0
TRCD : 2
TRAS : 5
TRP : 2
TRC : 7
TRFC : 16
TRRD : 3
TWR : 3
TWTR : 1
TRWT : 3
TREF : 3072 cycles
TWCL : Auto
Bank Interleave : Enabled
DQS Skew Control : Auto
DQS Skew Value : 0
Dram Drive Strength : Level 6
Dram Data Drive Str. : Level 4
Max Async Lat : 7.0 ns
Read Preamble : 5.0 ns
IdleCycle Limit : 256 cycles
Dynamic Counter : disabled
R/W Queue Bypass : 8x
Bypass Max : 7x
32 Byte Granularity : Disable(8 Bursts)
Well i'm quiet new to the a64 game ... :rolleyes:
Thanks for your help
Joe
Edit : Ok i tried 3.3V & 3.4V and @ 230Mhz @ 2-2-2-8 i'm getting approx. 100 Errors/pass in memtest #5.
They all seem to be at the same adress (986.0 MB)
Edit 2 : switched dimms. now the error occurs @ 53.0 Mb
since i'm using 3.3V 220Mhz is not memtest stable anymore !?!?!?
Twinmos TMII or TMSP?
So many complaints..so few with those..better chances better luck with the Twinmos imo.
We are overclockers if you want a sure thing then buy something you want thats guaranteed..."Thats not overlocking"
Overclocking is buying something thats not guaranteed and you do your best to get it there! Now thats "Overclocking!"
Some like taking risks with thier money and some don't. As you know overclocking is more of an art than a science an getting great results is not easy. Boards, CPU's and RAM all vary. This is the real source of most problems.Quote:
Originally Posted by godsa
If the Twinmoss suits your needs, and you have better results than we did with the tray we bought for testing, then this is a good thing.