***The Great Gskill 10 GIG HX PC 3200 Review!***

[fareastgq]

Well guys, it's been a while since I did the 3 gig redline review for everyone, so... I thought I would bring another great look at perhaps the most underrated ram on this forum.. Gskill HX. That's right folks, HX, not ZX or that other junk, this is gskills top of the line sticks. Is it worth it? we'll see.... hehehehe. This is what happens when u get bored and curious... U do things on a simple whim,...PIX WILL BE BELOW AT THE BOTTOM OF THE POST IF U WANNA GO THERE FIRST TO LOOK AT THEM
What we are going to answer here is :

1. Are they worth it over the cheaper ZX?
2. What are these since nobody has them?
3. What timings can these do? (prepare yourselves )
4. What will I get if I buy a pair for max fsb??
5. Is running 4 gigs really better than 2 gigs as far as gaming/apps goes?
6. Of course, this wouldn't be XtremeSystems without a max volts/mhz test!
7. What are you going to do with the rest u don't keep? (this I can answer right away too, like last time, they will be matched into pairs and put up in the FS section and sold locally)
8. Do you work for Gskill? (well this one, right away, I can say no, just like last time, I didn't work for mushkin either, just simply bored....)


This is the setup :
CBBWE 0550 TPMW opty 165 on stock opty cooling
4 gigs/2gigs/1gig of HX pc3200
DFI Ultra-D (ADO revison) bios 6-23
Gigabyte 6600 128mb PCI-E card @ stock
75 gig raptor
PCP&C PSU
XP pro corp Ed. SP2 (NO TWEAKS, fresh install)

Testing Methodology : 10 1 gig sticks of HX pc3200

-We will be using bios 6-23 (this is the stock one that came with the ADO board btw). LDT/FSB ratio will be set at x3, CPU multi = x9,
-All ram timings will be set to auto except for the main settings, ie: cas/ras to cas/ras precharge/Tras, better known as 2 3 2 5 for this particular ram at stock
-we will test tightest timings, loosest timings, highest FSB achievable memtest stable.
-We'll also be using a timeless classic 3dmark2001 to measure gaming performance and stability.
-We'll also be using Sandra memory bandwith benchmark to measure performance.
-We'll also be using super pi, 1 full run of all the tests at max fsb possible for 4 gigs 2t & 2 gigs 1t/2t (3x12 runs I believe that will be)
-Lastly, we'll use warcraft3 frozen throne, set on god mode with cheats enabled in a large map 12 group battle royal to see what is actually playable, IE doesn't crash,
-this will also enable us to directly see if 4 gigs vs 2 gigs vs 1 gig is any different.
-4 gigs will be ran in dual channel at 2T
-2 gigs will be ran in dual channel at 2T & 1T
-1 gig will be ran in single channel at 1T


Well, enuff of the details, lets get to the good stuff.

Here's what 10 sticks did right out of the box with 2.6 stock volts. this is MAX FSB MEMTEST 5 AND 8 stable.
1 ...280... (282 after 24 hour burn in with test 5)
2 ...274...(276 after 24 hour burn in with test 5)
3 ...272
4 ...271
5 ...269
6 ...268
7 ...263
8 ...263
9 ...262
10 ...254

Needless to say, this makes for some very nice matched pairs, with the exception of the last stick, they all hit 260+ fsb, realitively speaking when compared to other ram, 254 isn't too shabby either though. Also looks like burn in works on these sticks judging from the initial burn in with the 2 best sticks, looks like they are good for at least 2-5 mhz more, I would say an avg of 3 mhz gain if I were to make a guess at this point, time will tell...What does this equate to for you?? Pretty damned good 4/5th's chance that the pair u buy will clock 265 or over, 3/5ths chance that they will do 270 or over and a 1/10th chance that you will get an exceptionally good stick 280+!!!!

Here's data on 4gig dual channel 2T performance.
\ = test wasn't ran
n/a = was ran, not sucesssful
n/a with stats = was ran, not considered stable
Red = max fsb at tightest timings 3dmark/sandra stable

CPU Mhz---MULTIxFSB---Timings---3D MarK-----Sandra----Volts--Gigs

1.8--------9x200--------2 3 2 5----14,261-----4471/4484---2.6---4
1.953------9x217--------2 3 2 5----14882-----4818/4838----2.6---4
1.953------9x217------2.5 3 2 5----14823-----4786/4799----2.6---4
1.953------9x217--------3 3 2 5----14741-----4768/4783----2.6---4
2070-------9x230------2.5 3 2 5----15357-----5104/5097----2.6---4
2070-------9x230--------3 3 2 5----15177-----5057/5058----2.6---4
2160-------9x240------2.5 3 2 5----15637-----5306/5307----2.6---4
2160-------9x240--------3 3 2 5----15554-----5268/5272----2.6---4
2205-------9x245------2.5 3 2 5----15789-----5430/5433----2.6---4
2205-------9x245--------3 3 2 5----15704-----5388/5381----2.6---4
2295-------9x255--------3 3 2 5----15941-----5602/5622----2.6---4
2304-------9x256--------3 3 2 5----15998-----5617/5620----2.6---4
2313-------9x257--------3 3 2 5----16012-----5643/5629----2.6---4
2322-------9x258--------3 3 2 5----15469-----5653/5675----2.6---4
2322-------9x258--------3 3 3 5----16051-----5674/5668----2.6---4
2340-------9x260--------3 3 2 5------n/a------5712/5719----2.6---4
2430-------9x270--------3 3 2 5------n/a------5914/5926----2.6---4
2430-------9x270--------3 3 3 5------n/a----------n/a-------2.6---4

Here's data on 2gig dual channel 2T performance.
\ = test wasn't ran
n/a = was ran, not sucesssful
n/a with stats = was ran, not considered stable
Red = max fsb at tightest timings 3dmark/sandra stable
Purple = best of previous set of data


CPU Mhz---MULTIxFSB---Timings---3D MarK-----Sandra----Volts--Gigs

1.8--------9x200--------2 3 2 5----14,259 ----4480/4495---2.6----2
1.953------9x217--------2 3 2 5----14823-----4835/4847---2.6----2
1.953------9x217------2.5 3 2 5-------\-----------\--------2.6----2
1.953------9x217--------3 3 2 5-------\-----------\--------2.6----2
2070------9x230-------2.5 3 2 5----15270-----5103/5105---2.6----2
2070------9x230---------3 3 2 5-------\-----------\--------2.6----2
2160------9x240-------2.5 3 2 5----15554-----5310/5338---2.6----2
2160------9x240---------3 3 2 5-------\-----------\--------2.6----2
2205------9x245-------2.5 3 2 5----15695-----5442/5440---2.6----2
2205------9x245---------3 3 2 5-------\-----------\--------2.6----2
2295------9x255---------3 3 2 5 ----16018----5594/5595---2.6----2
2304------9x256---------3 3 2 5-------\-----------\--------2.6----2
2313------9x257---------3 3 2 5-------\------------\-------2.6----2
2322------9x258---------3 3 2 5-----16068----5642/5644---2.6----2
2322------9x258---------3 3 3 5-------\-----------\--------2.6----2
2340------9x260---------3 3 2 5----14940-----5709/5702---2.6----2
2340------9x260---------3 3 3 5-------\-----------\--------2.6----2
2385------9x265---------3 3 2 5----16216----5801/5808----2.6----2
2421------9x269---------3 3 2 5----16294----5888/5874----2.6----2
2430------9x270---------3 3 2 5--16352*n/a--5897/5913----2.6---2
2430------9x270---------3 3 3 5--16062*n/a------\---------2.6---2
2466------9x274---------3 3 3 5------n/a----------\--------2.6---2
2475------9x275---------3 3 2 5------n/a-----6042/6016----2.6---2
2484------9x276---------3 3 2 5------n/a-----6054/6076----2.6---2
2493------9x277---------3 3 2 5------n/a-----6081/6097----2.6---2

Here's data on 2gig dual channel 1T performance.
\ = test wasn't ran
n/a = was ran, not sucesssful
n/a with stats = was ran, not considered stable
Red = max fsb at tightest timings 3dmark/sandra stable
Purple = best of previous set of data


CPU Mhz---MULTIxFSB---Timings-----3D MarK-----Sandra----Volts--Gigs

1.8---------9x200 --------2 3 2 5----14,457------4888/4832--2.6---2
1.953------9x217---------2 3 2 5----15056-------5264/5206--2.6---2
1.953------9x217-------2.5 3 2 5-------\-------------\-------2.6---2
1.953------9x217---------3 3 2 5----15010-----------\-------2.6---2
2070------9x230--------2.5 3 2 5----15483-------5553/5488--2.6---2
2070------9x230----------3 3 2 5-------\-------------\-------2.6---2
2160------9x240--------2.5 3 2 5----15807-------5799/5749--2.6---2
2160------9x240----------3 3 2 5-------\-------------\-------2.6---2
2205------9x245--------2.5 3 2 5----15985-------5938/5858--2.6---2
2205------9x245----------3 3 2 5-------\-------------\-------2.6---2
2295------9x255----------3 3 2 5----16132-------6108/6036--2.6---2
2304------9x256----------3 3 2 5------\--------------\-------2.6---2
2313------9x257----------3 3 2 5-------\-------------\-------2.6---2
2322------9x258----------3 3 2 5----16160-------6168/6101--2.6---2
2322------9x259----------3 3 3 5-------\-------------\-------2.6---2
2340------9x260----------3 3 2 5-------\-------------\-------2.6---2
2340------9x260----------3 3 3 5----16256-----------\-------2.6---2
2385------9x265----------3 3 2 5----16424-------6346/6270--2.6---2
2394------9x266----------3 3 2 5----16313-------6352/6271--2.6---2
2403------9x267----------3 3 2 5----16436-------6365/6300--2.6---2
2412------9x268----------3 3 2 5----16401-------6411/6356--2.6---2
2421------9x269----------3 3 2 5----16409-------6426/6359--2.6---2
2430------9x270----------3 3 2 5----14621-------6551/6485--2.6---2
2430------9x270----------3 3 3 5----13544-----------\-------2.6---2
2466------9x274----------3 3 3 5------n/a------------\-------2.6---2
2475------9x275----------3 3 2 5------n/a-------6576/6517--2.6---2
2484------9x276----------3 3 2 5------n/a-------6605/6511--2.6---2
2493------9x277----------3 3 2 5------n/a-------6633/6553--2.6---2

Here's data on 1gig SINGLE channel 1T performance.
\ = test wasn't ran
n/a = was ran, not sucesssful
n/a with stats = was ran, not considered stable
Red = max fsb at tightest timings 3dmark/sandra stable
Purple = best of previous set of data


CPU Mhz---MULTIxFSB---Timings-----3D MarK-----Sandra----Volts--Gigs

1.8--------9x200----------2 3 2 5-----14,110----3084/3088---2.6----1
1.953------9x217----------2 3 2 5-----14630--------\---------2.6----1
1.953------9x217--------2.5 3 2 5--------\----------\---------2.6 ---1
1.953------9x217----------3 3 2 5-----14596-----3340/3338---2.6----1
2070-------9x230--------2.5 3 2 5-----15090---------\--------2.6----1
2070-------9x230----------3 3 2 5-------\------------\--------2.6 ----1
2160-------9x240--------2.5 3 2 5-----15499----------\--------2.6----1
2160-------9x240----------3 3 2 5-------\-------------\--------2.6----1
2205-------9x245--------2.5 3 2 5-----15603-----3758/3761----2.6----1
2205-------9x245----------3 3 2 5-------\-------------\--------2.6----1
2295-------9x255----------3 3 2 5-----15789----------\--------2.6----1
2304-------9x256----------3 3 2 5--------\------------\--------2.6----1
2313-------9x257----------3 3 2 5--------\------------\--------2.6----1
2322-------9x258----------3 3 2 5-----15949-----3954/3962----2.6----1
2322-------9x258----------3 3 3 5--------\------------\--------2.6----1
2340-------9x260----------3 3 2 5--------\------------\--------2.6----1
2340-------9x260----------3 3 3 5--------\------------\--------2.6----1
2385-------9x265----------3 3 2 5-----16110----------\--------2.6----1
2394-------9x266----------3 3 2 5-----16070----------\--------2.6----1
2403-------9x267----------3 3 2 5-----16026----------\--------2.6----1
2412-------9x268----------3 3 2 5-----16065----------\--------2.6----1
2421-------9x269----------3 3 2 5-----16128----------\--------2.6----1
2430-------9x270----------3 3 2 5-----16242-----4150/4150----2.6----1
2430-------9x270----------3 3 3 5--------\-----------\---------2.6----1
2439-------9x271----------3 3 2 5-----16273-----4161/4160 ----2.6----1
2448-------9x272----------3 3 2 5-----15739-----4180/4177----2.6----1
2457-------9x273----------3 3 2 5---14240*n/a---4191/4191----2.6----1
2466-------9x274----------3 3 3 5-----n/a--------4213/4207----2.6----1
2475-------9x275----------3 3 2 5--------\-------4217/4223----2.6----1
2484-------9x276----------3 3 2 5--------\-------4236/4238----2.6----1
2493-------9x277----------3 3 2 5--------\-------4260/4256----2.6----1
2502-------9x278----------3 3 2 5-------- \-------4268/4271----2.6----1
2511-------9x279----------3 3 2 5--------\-------4287/4288 ----2.6----1
2520-------9x280----------3 3 2 5--------\-------4301/4300----2.6----1
2529-------9x281----------3 3 2 5-------- \-------4312/4315----2.6----1
2538-------9x282----------3 3 2 5--------\-----------n/a-------2.6----1


Super Pi Runs @ 9x200mhz (2 3 2 5)
--------4 gig 2t------2gig 2t--------2gig 1t

32m----40m14.297----38m58.594----38m10.312
16m----18m37.188----18m38.442----18m17.078
8m-----08m21.406----08m23.203----08m12.031
4m-----03m50.032----03m50.172----03m45.390
2m-----01m45.078----01m45.125----01m43.110
1m-----00m45.609----00m45.641----00m44.718
512k---00m19.860----00m19.890----00m19.625
256k---00m08.296----00m08.282----00m08.250
128k---00m03.531----00m03.516----00m03.515
64k----00m01.578----00m01.563----00m01.562
32k----00m00.688----00m00.718----00m00.687
16k----00m00.297----00m00.297----00m00.281

Super Pi Runs @ MAX FSB @ 2.6 Volts

9x264 (2 3 2 5) 2T/4G----9x277 (2 3 2 5) 1T/2G----9x277 (2 3 2 5) 1T/1G
32m----30m15.312-----------------28m28.016----------------(coming)
16m----14m27.312-----------------13m32.172----------------(coming)
8m-----06m29.140-----------------06m04.000----------------(coming)
4m-----02m58.297-----------------02m46.891----------------(coming)
2m-----01m21.484-----------------01m16.344----------------(coming)
1m-----00m35.313-----------------00m33.109----------------(coming)
512k---00m15.359-----------------00m14.438----------------(coming)
256k---00m06.375-----------------00m06.063----------------(coming)
128k---00m02.703-----------------00m02.563----------------(coming)
64k----00m01.203 -----------------00m01.141----------------(coming)
32k----00m00.531-----------------00m00.500----------------(coming)
16k----00m00.218-----------------00m00.219----------------(coming)

Super Pi Runs @ MAX FSB @ 2.72 Volts

-----2T/4G (above was max fsb)---9x277 (2 3 2 5) 1T/1G---9x278 (2 3 2 5) 1T/1G
32m----------N/A--------------------28m17.37------------------- (coming)
16m----------N/A--------------------13m30.750-------------------(coming)
8m-----------N/A--------------------06m02.453-------------------(coming)
4m-----------N/A--------------------02m46.000-------------------(coming)
2m-----------N/A--------------------01m16.125-------------------(coming)
1m-----------N/A--------------------00m33.016--------------------(coming)
512k---------N/A--------------------00m14.468--------------------(coming)
256k---------N/A--------------------00m06.032--------------------(coming)
128k---------N/A--------------------00m02.562--------------------(coming)
64k----------N/A--------------------00m01.141--------------------(coming)
32k----------N/A--------------------00m00.515--------------------(coming)
16k----------N/A--------------------00m00.219--------------------(coming)

Super Pi Runs @ MAX FSB using 2.7v
maxed cpu mhz possible for that fsb 9/10 divider
9x308 (2 3 2 5) 2T/4G (277fsb)
32m---------(coming)
16m---------(coming)
8m----------(coming)
4m----------(coming)
2m----------(coming)
1m----------(coming)
512k--------(coming)
256k--------(coming)
128k--------(coming)
64k---------(coming)
32k---------(coming)
16k---------(coming)

[bachus_anonym]

So, you're saying that your best 2GBHX is 3D stable @ 270MHz 5-2-3-3.0-1T 2.6v? That's very impressive
Can you see how does situation look like with SPi32M, with that best pair? Would it go further that 3D-stable clock? For 4x1GB, you're using best 4 out of 10, right?
Awesome "legwork", man!

[fareastgq]

yeah, using the best 4 sticks for that 4 gig test. I'm sure it (the 2gig set) will go further, I'm hoping for 275 3dstable. Yeah, this ram is just freaking amazing, I haven't even thrown volts into the equation yet or tweaked any other timings. Yeah, the best was 270, I'm anxious to jump right to the end, but must resist... heh. I will say, one characteristic about these chips, they are weird, either your memtest stable, or u lock up or get a wierd screen, it doesn't error out much at all. Ok, I'll throw in a complete set of super pi runs too

[GSKILL TECH]

well i love the way you test and i will do my test like this later
i am trying to solve 3D problem also

[s7e9h3n]

Incredible work here ! I can't imagine being patient enough to test all these sticks

BTW, you need to edit question #8

8. Do you work for Gskill? (well this one, right away, I can say no, just like last time, I didn't work for mushkin either, just simply bored...and rich .)
[/COLOR]

[EnJoY]

Dude, awesome testing here! Thanks for doing this, although I'm sure it was a lot of fun, it looks like a lot of work too. More of us need to dispose of our income in more meaningful ways such as this.

[Markus_d]

nice results there!

and great work done!

pump some volts now and let the people here go crazy about the scores

[chinkgai]

what are the fates of the 8 sticks you arent going to keep? lol

[Kaiser_Sose]

Whats under the hood in these [ ICs ]?

[Major]

Wow ! thanks sharing all the effort with the community !

I can't imagine how much time went into all this testing, good results too !

[ozzimark]

Whats under the hood in these [ ICs ]?

infineon iirc.. whether it's B or C rev, i don't know. i'm guessing the latter from the scores

[Cpt Twitchy]

Very nice review and an astonishing amount of ram =O I wish I could play with all those goods.

[fareastgq]

what are the fates of the 8 sticks you arent going to keep? lol

Gonna throw them in FS of course,

BTW, you need to edit question #8

Laf... I wish I was.

Whats under the hood in these [ ICs ]?

I have no idea yet and since I don't wanna void the warranty, I don't think ppl want voided sticks so I don't think I'm gonna find out unfortunately.., I might pop off the spreaders on that last stick that did 254 if it doesn't get picked up though . Ozzy might be right though, if someone picks some up, feel free to void your warranty for us Thanks guys, I got 1 gig results coming, not exciting, but informative.... hehe.

[Bloody]

Howdy,

I got a pair of 2GBHX in my box and i couldn't believe what these RAMs can take... i got them running at stock 2.65v, clocked 266MHz @ 2.5-3-2-0 on a DFI Lanparty Expert, without extra chipset voltage or other modifications like drive strength and so on... Might even be a few MHz more but my CPU ain't the toughest draw so i don't have endless combinations there...

Who needs 300 MHz ?

[Vapor]

...3D stable?

[Bloody]

Yep, 9 hours Prime stable, then reinstalled Windows with all drivers etc, ran 3DMark05/06, Aquamark and then kicked some monster's asses in Sam2 for about 2 hours or so. Absolute stable. Linux also running fine..

Interesting is that 278MHz @ 2.8v / 3-4-4-8 was nogo. But however, i'm just fine with what i got, to say the least.

[Bloody]

Some more words..

266 MHz @ 2.5-3-2-0 / 2.65v:

*EDIT*: i was wrong with 266 as shown below in this thread, so:

250 MHz @ 2.5-3-2-0 / 2.65v:

Prime stable for about 9 hours (no errors), 3D stuff also stable,
incl. 3DMark05/06, Aquamark3 and some games, i.e. Serious Sam 2
played for 2 hours or so, C&C ZeroHour 8-player CompStomp
(5 hours, high CPU load as well as mem usage). The tRAS=0
doesn't gain anything but interesting that this is possible at all.

278 MHz @ 3.0-4-4-8 / 2.80v:

No way, couldn't even boot.

I didn't test if the RAMs run even better in the other two slots
(Lanparty Expert, slots 3/4 used). tRCD=2 was nogo, as expected, all other
settings still as-is (drive strength and so on).

Conclusion:

Note: the memory timings are displayed wrong because the screens were
made with a RAM setup for 233 MHz (228 * (7/6) = 266 MHz).

[biohead]

Bloody, I see you have an Expert. What happens when u lower volts to 2.5..2.4..2.3...?

http://www.xtremesystems.org/forums/...ad.php?t=88354

[fareastgq]

btw guys, I had no problems running any dividers at any combo, these 0550 chips have really nice controllers on them. I don't really think it has anything to do with the ram, I guess it could.

[ozzimark]

Note: the memory timings are displayed wrong because the screens were
made with a RAM setup for 233 MHz (228 * (7/6) = 266 MHz). I'm not even
sure about the eventual read throughput between CPU and RAM with this
setup (read speed seems a tad slow for my taste while write throughput
looks alright), but as mentioned above, the CPU doesn't allow me many
combinations. I'm facing a situation where my CPU can't keep up with
these RAMs..

hate to tell you, but if you're using an 11x multi and 7:6 ratio (233mhz), your ram will actually be running at cpu/10, for 250.87mhz

try your hand and see how far you can get with 1:1 with a lowered cpu multi

[Bloody]

@biohead: didn't test lower voltages yet, so much time spent in windows installation... but i guess i'll give it another test.

@ozzimark: thought i was fine because at 250 MHz 1:1, write speed
was lower at 250/1:1 Thanks for the note, of course i'll test this. Hang on..

[Bloody]

@ozzimark:

Unfortunately, you're right. A Prime test @ 256/1:1 failed after 8 minutes, so i'm a BIOS victim i guess..

Well... SORRY GUYS for the confusion about this.. Here's another shot from an 1:1 setup. (searching for an embarassment smilie)

[Bloody]

@biohead:

Just gave it a quick low-voltage test:

2.5v: 246MHz 2.5-3-2-0 SuperPi 1M O.K.
2.4v: 246MHz 2.5-3-2-0 SuperPi 1M O.K.
2.29v: 242MHz 2.5-3-2-0 SuperPi 1M O.K.

[fareastgq]

ok guys, updated with 1 gig single channel results. yeah, they are sucky, hehe. 36 runs of Super PI up next! We are getting to the volts, hahaha.

[fareastgq]

ok guys, updated with 72 runs of super pi! I got the max mhz/fsb tests in progress right now, to keep it all relevant, I did the max mhz test at the max fsb for an apples to apples comparison (after all, the ram is the focus here). I'll do an outright max mhz possible when I get back on water again so we can all see the diff. (and of course I will post screen shots, hahaha)
Some ram charateristics :

> 2.8v and up will not post
> 2.7v is the optimal volts
> setting volts at 2.73 actually hurt stability.
> LDT volts DO help stability and Max FSB
> I gained 1 fsb going from 2.6 to 2.7 (we are talking runs of 32m super pi here, not 1 min runs, I only ran the whole set if super pi 32 was able to complete)
> test 8 is useless for this ram (test 5 errors out, test 8 doesn't, good thing I always run test 5 first)
> running the 9/10 divider allowed me to boot into windows @ 280+, could not do this 1:1 (I'm still trying to figure this part out)