PDA

View Full Version : Who would be interested in large TCCD roundup on A64 Socket 939


JNav89GT
11-18-2004, 01:10 PM
I am in process of collecting samples from various manufacturers for high performance DDR to be reviewed in a roundup with Socket 939 Athlon64 hardware.
Platform will be tested to perform well at or above 275mhz. Planning on modifiying VDD on MSI K8N Neo2, and I have 2-3 boards to choose from.

In any event, here are samples I have for testing and those I am due to get. Also is list I plan to contact for samples. Please feel free to mention suggestions.

Have
G.Skill PC4400LE
G.Skill PC4400LD(supposed to be shipped this week)
OCZ PC3200 EL rev2
Confirmed coming
PQI 3200 2225(shipping)
Centon PC3200 2225 TCCD
PDP(spoke with Garret, stated they would send ram)

Probable
Geil Ultra X(waiting for approval from headquarters)
Mushkin PC3200 Level II V2(spoke with Tom, states they are in process of approving)




what do you guys think
testing will be done on Athlon FX55@3Ghz+ which should provide some very high bandwidth and hopefully better memory controller than run of the mill Winchesters. I will also include some Winchester data, due to requests of users in this thread. Main focus on article will be ultimate performance, so most benches will be FX related, but I will give some numbers with a Winchester core 3500+, and experiences with each ram on that system as well. Thanks everyone for ideas.

I'm up for ideas.
Article to be published at www.madshrimps.be
thanks
John
edit: updated list! Also, clarified cpu speed ;)
adding ocz 3200 back to roundup
sorry for confusion

Arkangyl
11-18-2004, 01:12 PM
I'd look forward to reading it

agenda2005
11-18-2004, 01:19 PM
Please include A64 3500+ NC for comparism purpose. I'll also be looking forward to seeing the numbers. Are you going to publish them in madshrimp?

jinu117
11-18-2004, 01:25 PM
3500 Newcastle might not cut it for these RAM. Mine will let up to 281mhz dual channel but found the 2 pair up to 285mhz and 290mhz on FX-53 I tried. Way to CPU dependent now.

G|-|oST
11-18-2004, 01:27 PM
This is a great idea! :toast: A follow up to the winnie article!

Winchester results would be awesome to see also.
Us poor folks would like to know :)

Good timing for such a write up too, a lot of people are
after some TCCD but not really sure whats best atm.

JNav89GT
11-18-2004, 01:32 PM
3500 Newcastle might not cut it for these RAM. Mine will let up to 281mhz dual channel but found the 2 pair up to 285mhz and 290mhz on FX-53 I tried. Way to CPU dependent now.
this is why I will probably use FX55, due to limitation of the winchester chips.
However, I will probably include a section on how each ram clocks in a 3500+ 0.09u setup. Again I will vmod that board.

I may have to pick up another FX55 to have better chance at high clocking mem controller :O :rolleyes:

JNav89GT
11-18-2004, 01:33 PM
Please include A64 3500+ NC for comparism purpose. I'll also be looking forward to seeing the numbers. Are you going to publish them in madshrimp?
yes at www.madshrimps.be

Stinger
11-18-2004, 01:38 PM
JNav89GT--> after your last article I will look forward to that cos I think the last was very very good.

naelq
11-18-2004, 02:16 PM
for sure

Blue078
11-18-2004, 02:59 PM
I would like read it. Sounds like a darn good idea

jmke
11-18-2004, 03:06 PM
you got my support ;)

trans am
11-18-2004, 03:10 PM
I have ocz el platinum 4200 and the pdp xbl if you want to borrow for the review.

Geforce4ti4200
11-18-2004, 03:19 PM
any chance of also including a winchester? Way more people have winchesters than FX and itll be interesting to see how much higher it clocks on the fx. If the ondie controller is the limit, some or all of that tccd will max out at whatever the ondie controller handles. This is the main reason I am extremely reluctant to go tccd when I pay an arm and I cant even use the ram to full potentional :mad:

jmke
11-18-2004, 03:19 PM
Hi Trans Am, the sneaky Asus review mailer ;)

Cranox
11-18-2004, 03:41 PM
Looking forward to it JNav89GT :up:

agenda2005
11-18-2004, 03:44 PM
If possible can you include benchmarks of EVEREST memory latency at different memory speed and timmings. You can also use other latency tst programs like cachemen/AIDA or whatever you can find around. It will provide a good comparison to see how memory latency changes with speed/timing.

Xerxes
11-18-2004, 04:01 PM
sounds awesome, i cant wait to see the results. even though i already bought some pqi (waiting on a cross ship though the first set was defective :/)

JNav89GT
11-18-2004, 04:03 PM
sounds awesome, i cant wait to see the results. even though i already some pqi (waiting on a cross ship though the first set was defective :/)
please let me know how you found their service

JNav89GT
11-18-2004, 04:04 PM
I have ocz el platinum 4200 and the pdp xbl if you want to borrow for the review.
ty for offer. I prefer not to borrow memory, as if something happens I don't want to be responsible. I will leave it up to OCZ and PDP if they choose to participate or not.

thanks though :)

Rabbi_NZ
11-18-2004, 04:33 PM
Great idea JNav89GT, look forward to seeing your review!

bacon
11-18-2004, 04:42 PM
great idea, especially if OCZ take part

JNav89GT
11-18-2004, 04:52 PM
plan is to start data in about 2-3 weeks
look for a review in about one month
I know it's awhile, but these things take time to put together, perform, and writeup :)
please be patient :)

althes
11-18-2004, 05:17 PM
look foward to it

mad mikee
11-18-2004, 07:40 PM
Sure to be interesting!! :banana4: :banana: :toast: :slobber:

Xerxes
11-18-2004, 10:39 PM
well JNav89GT im cross shipping with newegg since i just bought the ram. but pqi's customer service was great they responded promptly to my emails (way quicker than kingston which is the only ram manufacturer ive had personal and not business experience with) and were more than willing to give me a rma number the same day. So unless i get another bad batch i plan to stick with them. I get the new set monday probably and will give you an update on the results with them.

YoupY
11-19-2004, 03:12 AM
Can't wait to read it :D

Too bad it's gonna take a month, but I rather read a good review with lots of usefull data then one that's quickly put together and doesn't have what it takes to be a good review.
I hope you can get every reseller to cooperate.

Good luck and have fun writing it :D

JNav89GT
11-19-2004, 07:03 AM
How would you like a section after benching that lists MHz per $$$. Not sure about this though. My feeling this though as partially, oc'ing A64 is luck of controller, luck of good board, luck of getting ram that runs higher than rest of field.

Xerxes
11-22-2004, 02:39 PM
a little update for you JNav89GT got my replacement pqi and its running great. has passed test 5 in memtest several times so far and it even posted 2-2-2-5 at 2.55v the old batch wouldn't even do that.

JNav89GT
11-22-2004, 02:55 PM
well PDP confirmed :)

Still haven't heard from OCZ or Adata :(

burnhead
11-23-2004, 06:03 AM
really interested...

iboomalot
11-23-2004, 07:32 AM
One thing I would like to see is actual time savings.

example lets say you run two systems

both clocked at 2.7 ghz but different ram timings and FSB speed

system 1 = 300*9 = 2.7ghz ram timings = 3-4-4-8
system 2 = 246*11=2.7ghz ram timings = 2-2-2-5

system 1 looks to be faster due to more memory bandwidth(fsb) but with the higher timings which system is FASTER and takes less time to finish a project.

some benches showing actual speed of the rams would be nice.

thanks for the tests in advance

jmke
11-23-2004, 07:37 AM
no off the shelve memory will run 246 in DC 2-2-2-5 :)

iboomalot
11-23-2004, 10:09 AM
BH5 will with proper voltages and that was mearly an example :D

JNav89GT
11-23-2004, 10:24 AM
One thing I would like to see is actual time savings.

example lets say you run two systems

both clocked at 2.7 ghz but different ram timings and FSB speed

system 1 = 300*9 = 2.7ghz ram timings = 3-4-4-8
system 2 = 246*11=2.7ghz ram timings = 2-2-2-5

system 1 looks to be faster due to more memory bandwidth(fsb) but with the higher timings which system is FASTER and takes less time to finish a project.

some benches showing actual speed of the rams would be nice.

thanks for the tests in advance
I agree that would be interesting, however I need to concentrate on currently available products that are in channel and specifically 1GB kits. I had 10kits+ of Mushkin PC3500 Level II and only 2-3 of them would go over 250+mhz even with 3.3v.

Ubermann
11-23-2004, 10:47 AM
Off topic, there is a new sandra 2005 out.
But it still cant detect T1 or T2, annoying.
I can make some tests with different timings if you want.

iboomalot
11-23-2004, 12:55 PM
I agree that would be interesting, however I need to concentrate on currently available products that are in channel and specifically 1GB kits. I had 10kits+ of Mushkin PC3500 Level II and only 2-3 of them would go over 250+mhz even with 3.3v.

maybe I should have used a better example to keep it less confusing
------------------------------------------------------------------------
both clocked at 2.7 ghz but different ram timings and FSB speed

system 1 = 300*9 = 2.7ghz ram timings = 3-4-4-8
system 2 = 246*11=2.7ghz ram timings = 2.5-2-2-8

might even be the same RAM just like to see real world results testing is all.


just because one ram might have a higher bandwidth on sandra yet be slower overall due to timings.

JNav89GT
11-23-2004, 01:00 PM
i see your point, but I really doubt these modules will go to 240+ at 2.5 2:2:5
maybe if I could get some Ballistix PC4000 but I'm knee deep in a Ballistix DDR2 review and couldn't ask for new samples till I was done here, plus I think they are on to promoting newer memories.

iboomalot
11-23-2004, 01:04 PM
jesus christ people

I could care less if the timings are 5-4-5-20 or 1-1-1-5

its ONLY AN EXAMPLE !!!!!!!!!!!!!!!!!!

It would just be nice to see the TOP or FASTEST useable speed and timings on the ram you are testing.

Kinda like they do on anandtech , just because you can clock a stick to DDR600 doesn't make it good if the timings are slop.

thanks for doing the test that is very nice of you :)

JNav89GT
11-23-2004, 01:08 PM
well in that case, I generally do that anyway.
Please refer to my new corsair article
while this isn't as stratified as you would like, I will keep this in mind for my roundup.
However, I agree, keeping cpu speed constant is important and allowing timings/HTT to illustrate performance differences.

iboomalot
11-24-2004, 07:40 AM
cirsair link please :)

JNav89GT
11-24-2004, 07:40 AM
http://www.madshrimps.be/?action=getarticle&articID=254

Arnar
11-24-2004, 07:43 AM
Thats sounds really nice ;)

TEDY
11-24-2004, 09:49 AM
no KHX3200ULK2/1G ?

i got this but live in slovenija...just got them today

Holst
11-24-2004, 09:55 AM
Looking forward to seeing the results.

iboomalot
11-24-2004, 11:07 AM
nice review

I like how it shows the difference the timings lower the gaming and bandwidth.

JNav89GT
11-25-2004, 03:13 PM
list updated

Liquid3D
11-25-2004, 10:29 PM
well PDP confirmed :)

Still haven't heard from OCZ or Adata :(

You were offered OCZ PC4200 from TransAm? :confused:

And by the way Trans, that was a very kind offer. :D

Liquid3D
11-25-2004, 11:01 PM
no off the shelve memory will run 246 in DC 2-2-2-5 :)

What do you mean "off the shelve" ?

With all due respect boss, if a memory brand pulled "off the shelve" is included in a round-up and can't perform at such and such a speed, yet the other brands which were sent specifically for that round-up could, I would name that round-up "Manufacturer's who hand-pick Review Samples." Often time when people ask me why I don't do round-ups all that often, it's because some companies justifiably won't subject their memory knowing some manufacturer's do hand-pick.

I've got samples from Kingston, Corsair, and OCZ which wouldn't OC above their rated speeds, and while I was dissapointed, and one company was angry at me, at the very least I knew they weren't "hand-picked." In so far as the maker's mentioned Corsair and Kingston are very adamant about HOW their memory is reviewed. I don't blame them, I've learned a few things over the years, honesty is what gets you somewhere in this business. At present there's only two brands of DDR2 I use in my system right now, Kingston Hyper-X, and Corsair Twin-2X.

If you mean "off the shelve" as in lesser known brands, then we should remember it's the ADATA's of the industry which often surprise us. Since there's only one or two Fab's on the planet making these chips, and a few more making PCB's, sometimes "off the shelve" is your best bet.

The entire premise, history, of Overclocking was begun by the more experienced enthusiasts more ofte then not, on a budget. Those hobbyists whom knew how to find those "off the shelve" runs which were highly "overclockable." Most of the time that inside info was carefully gaurded because it was hard earned through research, and investment in "potentially overclockable" parts. Anyone with a lot of money can overclock, but overclocking was begun by the PC-enthusiast on a budget looking to get more for his/her dollar. It was also nurtured by those whom have a love for the technology, and what it can do. I'm of the latter two. I'd much rather be labled an Enthusiast, then an Overclocker.

saaya
11-26-2004, 10:28 AM
def go up to 3.3v at least and if possible remove the heatspreaders to compare the week steppings to see if theres an indication in the manufacturing date and ocability and how well it reacts to vdimm and how well it runs with 2-2-2 :D

JNav89GT
11-26-2004, 10:32 AM
I will not be able to crank voltages or remove spreaders for the purposes of this review. The review needs to be kept in mainstream context, not just for us here in the lunatic fringe :).

jmke
11-27-2004, 11:08 AM
What do you mean "off the shelve" ?
.

in mean exactly what I wrote down sir. You can not buy memory now which runs those timings. simple as that.

saaya
11-27-2004, 11:52 AM
hmmm i think keith brought up a good point! you guys have good relations to distros, right? try to get retail memory for the test and dont use any "review samples" from manufacturers, not only anandtech has been getting "surprisingly good results" with memory that performed a lot worse for most people who bought the memory later...

im not saying any memory company cheated, i think they probably get some few ics first to test the quality, and then build some test sticks and see how they like it, those test ic's are probably better clocking than the average ic they will later buy, and i guess they sent one of the early testing sticks to reviewers wich had good ic's on it and performed better than the average stick in retail.

anmyways, if you get the sticks in retail you can be 100% sure that this wont happen, so its always a good thing to do if you have the option to do it :)

jmke
11-27-2004, 12:01 PM
I'm doing my TCCD roundup on S754 using memory from www.pc-memory-upgrade.co.uk , only Mushkin & G.Skill were sent by the companies; neither were hand picked as far I can tell from the results compared to others I've seen in the forums.


but the main problem is getting a retailer to agree for setting you up with the material; I got very luch with PC-Memory-Upgrade who have been a big help; but this rather the exception then the rule

saaya
11-27-2004, 12:03 PM
yeah i know... but its always best to use that possibility if you can...

pc memory upgrade co uk is a great shop! really nice people!

computersmsa
11-27-2004, 12:11 PM
It's good idea !
Don't forget ADATA PC4500 :) (Same than PC4800)

++

MaxxxRacer
11-27-2004, 02:11 PM
Jnav, you said you removed the ocz from the review? Are you going to put the new 3200 plat rev2's on the review. I would really like those to be on there as that is the ram i was planning on getting in about a month.

JNav89GT
11-27-2004, 02:14 PM
Yes I believe I will use them. :)
thanks