-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
The Unofficial Official Ballistix Results Thread
- Thread starter Odie812
- Start date
- Joined
- Feb 7, 2004
- Location
- Hessen, Germany
wfarid said:
But noone else managed to get results like anand did.
Most modules make 260MHz, and then hit this wall- I think BH-5 are still better Low latency sticks, if you can give them volts
- Joined
- May 30, 2004
- Location
- Pleasantville
OP
- Joined
- Oct 2, 2003
- Location
- Pennsylvania
- Thread Starter
- #85
Mark E. Smith said:
That's also what my sticks seem to do: 235mhz or so at 2.5-2-2-X. I also wasn't able to get much of an OC out of my sticks. I got to 270 1:1 but they were nowhere near stable with 2.8V. I could run 265 fine at 2.5-3-3-8 with a little blip now and again. I think if I had about 3.0V they would open up a little bit. I'm glad to see that we're all seeing similar capabilities of this memory.
I don't know about you all, but I'd take a solid, consistent memory module anyday. There seems to be consistency in the Ballistix line, and I really dig that, as opposed to, say, the Buffalo 5B -C disaster with the PC3700 sticks where some people were getting PC3200 and some were getting PC3700, some blue-screened, some didn't... you get the idea. I think Crucial has a fine product on their hands, both with the PC3200 and PC4000 flavors of Ballistix.
Sincerely,
Odie
- Joined
- May 30, 2004
- Location
- Pleasantville
Odie812 said:
I wonder why the micron chips get so damn hot. After memtesting for an hour or so the heatspreaders became barely untouchable. With 3.0V you should have active cooling.
ATM the database is too small to provide evidence for such consistency. Bigtoe mentioned that the first week Ballistix (reviewed at anandtech) performed much better than newer ones. It would be interesting to know how crucial is binning the chips. Are PC4000 Ballistix just better performing PC3200?
Ballistix of the same batch should perform equal - mine have been manufactured in week 32:
OP
- Joined
- Oct 2, 2003
- Location
- Pennsylvania
- Thread Starter
- #87
Yes, the heat is definitely an issue. Even with 2.7V my chips get pretty hot.
Where is that pic from?
Where is that pic from?
- Joined
- May 30, 2004
- Location
- Pleasantville
I just tested some 2x512 MB Ballistix PC3200 from Newegg. In a nutshell, the ram reminds me a lot of XMS3500 with BH-5, in that it is a *****cat with Memtest: very few errors, errors out consistently in duplicate runs, and nary a #6 error on the way up to max overclock. But it runs at waaay lower voltage. These are my results on a finicky IS7-E board which maxes out at 2.8 VDIMM:
All results are A-A-A-D-D unless indicated otherwise:
221, 1:1, 2-2-2-5 @ 2.6 volts
225, 1:1 2-2-2-5 @ 2.8 volts, Sr-Enh-A-D-D (!!!)
242, 1:1, 2.5-2-2-5 @ 2.6 volts
245, 1:1, 2.5-2-2-5 @ 2.7 volts
248, 1:1, 2.5-2-2-5 @ 2.8 volts
289, 5:4, 2.5-2-2-5 @ 2.6 volts
282, 5:4, 2-2-2-5 @ 2.8 volts, Sr-Enh-A-D-D (equivalent to 225, 1:1)
If you wonder why I did so much testing at 2.6 volts, I thought that was default voltage. I have since found out Crucial uses 2.8. Nevertheless, the results are impressive.
The PC3200 must be speed binned for fast timings down low, as the ram will not run relaxed timings w/o errors in Memtest. 2-3-2-5, 2-3-3-6, 2.5-3-3-6, 3-2-3-5, or 3-3-3-5 all don't get me higher, and the ram maxes out at 248 fsb, albeit at 2.5-2-2-5. It does seem I might get higher with >2.8 volts. With 2.85, I might do 250-252, not quite the 257 of the (cherry-picked) sticks in the Anandtech review. However my sticks are better down low, doing 221, 2-2-2-5 at just 2.6 volts, as well as 242, 2.5-2-2-5.
I was amazed to boot at Sr-Enh-A-D-D! But I cannot run these GAT timings at anything other than by 2-2-2-5. I think this is an Abit bug. The only other ram I have ever been able to run GAT on is my XMS3500 BH-5, at 2-2-2-5 too, of course.
You won't believe the unbuffered bandwidth of this Abit 865P board at a mere 225, 1:1, 2-2-2-5, Sr-Enh-A-D-D for full PAT: 3564/3582. I backed that up by running same theoretical fsb at 5:4, e.g., 283 fsb, 5:4, 2-2-2-5: 3549/3536. UNREAL! I have never seen such memory bandwidth at so low a fsb. 5:4 tested to be 3D stable too.
In regards to 5:4, the ram should theoretically do 303. 5:4, 2.5-2-2-5 @ 2.6 volts, however 3D stability was only achieved at 289 fsb. This is expected due to the fast 2.5-2-2-5 timings and the fact 5:4 stresses the board out more.
Normally an unmodded IS7/IC7 is strictly a 3:2 board due to the max 2.8 VDIMM and finicky nature of Abit boards with ram. But this Ballistix ram is just the ticket for these boards. Crucial has come a long way with ram, and really did a good job with these sticks. Only negative is the price of the ram: $138 per 512 MB stick.
As far as maxing out around 248 at 1:1, 255-260 at 3-3-3-5 would be nicer, but those ICs were probably put in the Ballistix PC4000 stuff. I might get higher on an AI7 which can go to 3.2 VDIMM, though other posts on this thread indicate max VDIMM the ram moves is 2.9. I can still get a killer rig on the IS7 by running a 3.2C at 248, 1:1, 2.5-2-2-5 @ 2.8 volts. That's 3.97 gig folks. May need phase change (but I got that). Or how about a 2.80C at 283, 5:4, 2-2-2-5, full PAT. 3.96 gig with 3550 unbuffered bandwidth at 2-2-2-5 . The 3DMarks will really love that.
I dare say posting the timings I achieved with this ram at such low voltage would rain on the parade of that other thread regarding the PQI ram with Sammy TCCD. The Sammy can't match the ram down low at 2-2-2-5 timings, particularly with such low voltage. To be fair, the PQI ram is waaay cheaper.
All results are A-A-A-D-D unless indicated otherwise:
221, 1:1, 2-2-2-5 @ 2.6 volts
225, 1:1 2-2-2-5 @ 2.8 volts, Sr-Enh-A-D-D (!!!)
242, 1:1, 2.5-2-2-5 @ 2.6 volts
245, 1:1, 2.5-2-2-5 @ 2.7 volts
248, 1:1, 2.5-2-2-5 @ 2.8 volts
289, 5:4, 2.5-2-2-5 @ 2.6 volts
282, 5:4, 2-2-2-5 @ 2.8 volts, Sr-Enh-A-D-D (equivalent to 225, 1:1)
If you wonder why I did so much testing at 2.6 volts, I thought that was default voltage
. I have since found out Crucial uses 2.8. Nevertheless, the results are impressive.The PC3200 must be speed binned for fast timings down low, as the ram will not run relaxed timings w/o errors in Memtest. 2-3-2-5, 2-3-3-6, 2.5-3-3-6, 3-2-3-5, or 3-3-3-5 all don't get me higher, and the ram maxes out at 248 fsb, albeit at 2.5-2-2-5. It does seem I might get higher with >2.8 volts. With 2.85, I might do 250-252, not quite the 257 of the (cherry-picked) sticks in the Anandtech review. However my sticks are better down low, doing 221, 2-2-2-5 at just 2.6 volts, as well as 242, 2.5-2-2-5.
I was amazed to boot at Sr-Enh-A-D-D! But I cannot run these GAT timings at anything other than by 2-2-2-5. I think this is an Abit bug. The only other ram I have ever been able to run GAT on is my XMS3500 BH-5, at 2-2-2-5 too, of course.
You won't believe the unbuffered bandwidth of this Abit 865P board at a mere 225, 1:1, 2-2-2-5, Sr-Enh-A-D-D for full PAT: 3564/3582. I backed that up by running same theoretical fsb at 5:4, e.g., 283 fsb, 5:4, 2-2-2-5: 3549/3536. UNREAL! I have never seen such memory bandwidth at so low a fsb. 5:4 tested to be 3D stable too.
In regards to 5:4, the ram should theoretically do 303. 5:4, 2.5-2-2-5 @ 2.6 volts
, however 3D stability was only achieved at 289 fsb. This is expected due to the fast 2.5-2-2-5 timings and the fact 5:4 stresses the board out more.Normally an unmodded IS7/IC7 is strictly a 3:2 board due to the max 2.8 VDIMM and finicky nature of Abit boards with ram. But this Ballistix ram is just the ticket for these boards. Crucial has come a long way with ram, and really did a good job with these sticks. Only negative is the price of the ram: $138 per 512 MB stick.
As far as maxing out around 248 at 1:1, 255-260 at 3-3-3-5 would be nicer, but those ICs were probably put in the Ballistix PC4000 stuff. I might get higher on an AI7 which can go to 3.2 VDIMM, though other posts on this thread indicate max VDIMM the ram moves is 2.9. I can still get a killer rig on the IS7 by running a 3.2C at 248, 1:1, 2.5-2-2-5 @ 2.8 volts. That's 3.97 gig folks. May need phase change (but I got that). Or how about a 2.80C at 283, 5:4, 2-2-2-5, full PAT. 3.96 gig with 3550 unbuffered bandwidth at 2-2-2-5
. The 3DMarks will really love that.I dare say posting the timings I achieved with this ram at such low voltage would rain on the parade of that other thread regarding the PQI ram with Sammy TCCD. The Sammy can't match the ram down low at 2-2-2-5 timings, particularly with such low voltage. To be fair, the PQI ram is waaay cheaper.
Last edited:
- Joined
- Feb 26, 2004
Very nice indeed!
As far as rainning on the parade, I think everyone is interested in having as many options to choose from, and this is certainly good information. Coming from you, I put much more stock in these results than from what may have been cherry-picked ram selected for a commercial review. The PQI price advantage may not hold, as newegg has a habit of adjusting their prices based on a product's demand, and they have sold out of 5121 sticks of the PQI. The next lowest priced place that has them in stock sells, get this, the Platinum Heatspreader variety for $262/gig pack and the "more desireable" Black Mirrored dual pack for $286. That's priced right there with the Ballistix. I just bought 3 gigs of ram this week and 5 1/2 over the last two weeks and have totally exhausted my ram buying budget for this year (and next???) otherwise I would have to go out and get me some of this stuff. Then again...anyone want some Hynix...cheap? (not a sale offer, just some kidding going on here).
I will place a link to this thread at the PQI to let those guys know just how well your Ballistix played and that this option exists.
As far as rainning on the parade, I think everyone is interested in having as many options to choose from, and this is certainly good information. Coming from you, I put much more stock in these results than from what may have been cherry-picked ram selected for a commercial review. The PQI price advantage may not hold, as newegg has a habit of adjusting their prices based on a product's demand, and they have sold out of 5121 sticks of the PQI. The next lowest priced place that has them in stock sells, get this, the Platinum Heatspreader variety for $262/gig pack and the "more desireable" Black Mirrored dual pack for $286. That's priced right there with the Ballistix. I just bought 3 gigs of ram this week and 5 1/2 over the last two weeks and have totally exhausted my ram buying budget for this year (and next???) otherwise I would have to go out and get me some of this stuff. Then again...anyone want some Hynix...cheap? (not a sale offer, just some kidding going on here).
I will place a link to this thread at the PQI to let those guys know just how well your Ballistix played and that this option exists.
I paid $158 each for my sticks from Newegg , when they had them in stock anyway. Rumor had it only the first batch week of 0426 (Anand sticks) were good, but my ram is 0432, same as Mark's. Hell, AMD guys would kill for these sticks. 242, 2.5-2-2-5 @ 2.6 volts! If you don't already know, 2.5-2-2-5 is only a tad slower than 2-2-2-5, less than 100 3DMark2001 points lower in the 15k area.
Morever the sticks were in different boxes. One had a Crucial label with 'ABS Computer Technologies' of Industry, CA on it, while the other had no label. The s/ns are way different in AIDA too. I've not tested the sticks individually.
The PCB layout of the Ballistix is very different from the Crucial PC3200 with -5B Cs. It's less busy with only two types of resistors (or whatever those rectangular thingies are). It's clear to me that Crucial has very carefully binned both the ICs and the sticks. Even if some other ram manufacturer wanted to buy the ICs to make their own sticks, I bet Crucial got the good stuff (charity starts at home, after all).
I found I could bump GAT settings from A-A-A-D-D to A-Str-A-A-A. Still checking bandwidth difference. I'm typing on the computer now @ 242, 1:1, 2.5-2-2-5 (2.6 VDIMM!) with those settings. Getting 3450 unbuffered or so, PAT disabled.
What I like about the Crucial ram is the 5:4 ability on these finicky Abit boards. This Ballistix stuff should run awesome on the Asus boards. Morever, I notice absolutely NO artifacts on the 865P IS7-E at either 1:1 or 5:4, something I did see on the -5B Cs from Buffalo/Crucial. I dunno whether it's luck or homework on Crucial's part.
, when they had them in stock anyway. Rumor had it only the first batch week of 0426 (Anand sticks) were good, but my ram is 0432, same as Mark's. Hell, AMD guys would kill for these sticks. 242, 2.5-2-2-5 @ 2.6 volts! If you don't already know, 2.5-2-2-5 is only a tad slower than 2-2-2-5, less than 100 3DMark2001 points lower in the 15k area.Morever the sticks were in different boxes. One had a Crucial label with 'ABS Computer Technologies' of Industry, CA on it, while the other had no label. The s/ns are way different in AIDA too. I've not tested the sticks individually.
The PCB layout of the Ballistix is very different from the Crucial PC3200 with -5B Cs. It's less busy with only two types of resistors (or whatever those rectangular thingies are). It's clear to me that Crucial has very carefully binned both the ICs and the sticks. Even if some other ram manufacturer wanted to buy the ICs to make their own sticks, I bet Crucial got the good stuff (charity starts at home, after all
).I found I could bump GAT settings from A-A-A-D-D to A-Str-A-A-A. Still checking bandwidth difference. I'm typing on the computer now @ 242, 1:1, 2.5-2-2-5 (2.6 VDIMM!) with those settings. Getting 3450 unbuffered or so, PAT disabled.
What I like about the Crucial ram is the 5:4 ability on these finicky Abit boards. This Ballistix stuff should run awesome on the Asus boards. Morever, I notice absolutely NO artifacts on the 865P IS7-E at either 1:1 or 5:4, something I did see on the -5B Cs from Buffalo/Crucial. I dunno whether it's luck or homework on Crucial's part.
Last edited:
I got some Crucial BallistiX PC3200 (2x512MB).
Currently I have them running at 290Mhz@3,0v (2,9v in Bios) with 3-3-3-5-1T. All 100% memtest86 stable. Didn't try to lower voltage, but perhaps it will still be stable at 2,9v or 300Mhz. Have to try that, but my CPU won't stay stable at 300x9 for now.
Really nice stuff. Before that, I had them maxed at 233Mhz@2,7v with 2,5-2-2-5-1T. Everything above isn't memtest86 stable anymore with 2,5 CAS.
Currently I have them running at 290Mhz@3,0v (2,9v in Bios) with 3-3-3-5-1T. All 100% memtest86 stable. Didn't try to lower voltage, but perhaps it will still be stable at 2,9v or 300Mhz. Have to try that, but my CPU won't stay stable at 300x9 for now.
Really nice stuff. Before that, I had them maxed at 233Mhz@2,7v with 2,5-2-2-5-1T. Everything above isn't memtest86 stable anymore with 2,5 CAS.
Last edited:
r4ph43l said:I got some Crucial BallistiX PC3200 (2x512MB).
Currently I have them running at 290Mhz@3,0v (2,9v in Bios) with 3-3-3-5-1T. All 100% memtest86 stable. Didn't try to lower voltage, but perhaps it will still be stable at 2,9v or 300Mhz. Have to try that, but my CPU won't stay stable at 300x9 for now.
Really nice stuff. Before that, I had them maxed at 233Mhz@2,7v with 2,5-2-2-5-1T. Everything above isn't memtest86 stable anymore with 2,5 CAS.
I still have my Ballistix and I did try them at 3-3-3- once on a DFI NF4 board with a CPU that does 300+, 2.5-3-3-7 with TCCD. No good at 3-3-3-, 1T. I think the highest they did was 270 or something. This is even though I can do 2.5-2-2-5 quite a bit higher than you (around 244-250). I'm talking 2x512s.
I know Ballistic loosened up things as time went on, in that you can run higher up with looser (and tighter as in 2.5-2-3-5) timings. Maybe you have the later sticks.
- Joined
- Jul 2, 2004
- Location
- Grand Rapids, MI.
r4ph43l said:
Awsome. I am getting higher bandwidth with 2.5-2-2-6 than you are with your -5b-C's. I can get 2.5-2-2-6 @ 268MHz 1T off with -5b-G's. Right now I am limited on HTT do to bad mobo, but in the near future I am getting the DFI with a 3700 Newark. Just makes me wonder what I can do on bandwidth.
Clevor said:
Yeah, I got my sticks only a month ago. Seems they have used better chips now. I'd try to remove the heatspreader to check the IC's but I'd rather like to keep warranty for now, since the sticky pads are really sticky (I pulled it quite something, but couldn't get it loose).
Yeah, bandwidth isn't that impressing for such timings, but I'm more than happy with that, as I paid 150€ for them and was worried that I should have bought the OCZ Gold when I noticed I was maxed at 233Mhz with 2,5CL. Could've been worseavalanche83 said:
Btw: had to get back down to 277x9 for now as my CPU wouldn't stay prime stable long enough on 290 (also had crashes in CS ~ bah). Even with 1,6+v. Maybe I need some more burning-in to max out more.
- Joined
- Jul 2, 2004
- Location
- Grand Rapids, MI.
Try lowering your multi instead of bringing down the HTT. Maybe the -5B-G squezzes out a little more on 2.5 
.
.Similar threads
