Jump to content


  • Content Count

  • Joined

  • Last visited

About YYZ2112

  • Rank
    New Member
  1. I did fail to mention in my earlier analysis that the exact same problems (system freeze) also occured when the ext. drives were connected to the integrated usb ports on the rear mobo i/o panel. This exact anomaly was reproduceable using: 1. 1394 on i/o panel USLogic 60gb drive via mobo header 2. 80 gb Western Digital usb 2.0 drive (i/o panel or mobo header) 3. 160 gb Seagate usb 2.0 drive (i/o panel or mobo header) Odds of three defective USB/1394 controllers on these ext. HD's when they work flawlessly on my other two machines are fairly low. Again, I have not upgraded the cooler on the chipset so I can't say for sure if that will solve my problem or not. I have also not implemented the hotfix that solved kapbot's issue. I'll be looking for some 'tinkering time' soon. Thx.
  2. Ok, first of all I always thought the OP was the Operator (or Admin type) so of course I missed the Original Poster reference. Sorry. There has been no change in the sub-par USB/1394 data transfer performance of my DFI machine since the last comments I made on this thread. My solution has been to simply do external HD transfers through my other two significantly 'weaker' machines (USB performance aside). I have not yet installed a better heatsink on my chipset simply because I have not had time to 'tinker' with my PC for a very long time. The heatsink solution still may work but since I have the other PC workaround, I haven't been compelled to do it. Sorry guys :/ Powersaving on my USB ports is disabled. My DFI machine has performed flawlessly since with the sole exception of the USB port data transfer problem. I hope this helps update the status of this problem. I also hope Ominous is still subscribed to this thread so he can say if his problem was fixed or not and it does seem that the 'hotfix' that kapbot used fixed his so maybe that's it. I'll update this thread when I find more time to tinker. Thanks.
  3. . I tested by simply moving files (large over 250mb maybe) to/from a 2.0 usb ext drive and you'll be able to tell the difference right off....if it takes more than say 25-30 seconds (at the most!) then you don't have high speed, otherwise you do. It's a huge difference than 1.0 so you'll know in a few seconds if it's enabled. I'm sure there is a more technical and 'correct' answer out there but this is my USB2.0-for-dummy's technique . GL! If it still doesn't seem to be fast enough, re-update the drivers but point to the DFI CD-ROM....as otherwise suggested. USB2.0 drivers are custom for your OS, not your hardware. There used to be modular MS USB2.0 drivers out there early on, but with the release of sp1, they removed the modular install package so they could compel the masses to upgrade. I now actually use an old ASUS mobo driver CD-ROM to install 2.0 drivers for my older XP v2600 installs (no service pack).
  4. Just a hunch but I believe if you go to your usb controllers in your hardware device manager USB properties (right click on MyComputer, hardware....etc. etc.) if you find the controller that says 'enhanced'....and update the driver for that, you will have enabled USB2.0. GL.
  5. I know for sure it is the DFI mobo now because I have 2 other PC's networked to my primary system (the one in my Sig). To get any data successfully transferred between the USB or 1394 ext HD's and to/from my DFI machine, I just hook them up to the other 2 PC's USB/1394 ports and control them (they're all shared HD's) from my primary DFI machine. Both networked machines have older NF2 chipset socket 462 XP3200+ and 2400+ and they handle maximum USB and 1394 throughput flawlessly, 1hr of constant data transfer and not a single hiccup! My DFI external data ports won't last 90 seconds! I'm betting I can solve the problem with a better HSF on my DFI mobo chip, but if DFI is going to put 10 USB2.0 and a 3194 port on the MOBO (I'm using nearly all of them), it should @#[email protected]'n well work with stock chipset cooling,...yea, sorry but this has got me a little torqued. Thanks for all the input guys. I might just end up hacking up one of my GeForce2 all-copper low profile Vid card HSF's I have laying around and AS-Adhesive it to the chipset. It'd be a lot beafier than the Evercool vc-re (which I did check out) and would cost me zilch. I don't mind voiding a warranty (stock HSF removal) because I want to improve performance, but doing so to simply get the product to work as spec'd is a bit hard to swallow. Oh, and BTW, CPU idles at around 28C (36 load) and I have plenty of airflow inside the case so there is no chance the problem is due to high case temps. Thanks again.
  6. Thanks for the suggestions. No, Ominous it doesn't bluescreen. When I'm copying or moving data between or to/from an external USB or 1394 drive, the window with the 'flying' folder just stutters (2/10's of a sec or so) and then the whole system locks up (freezes in place) no mouse control, no flying window animation, the HD activity LED's go dead (although the HD are of course still powered on). These lockups happen every time anywhere from 30 seconds to 90 seconds into the data transfer which leads me to believe it could be a heat problem. Again, this ONLY happens with data transfers involving external HD's. If anyone has an Infinity board with the stock HSF on the South(or is it North) bridge (south end of mobo east of vid card slot(s)) please test this scenario and let me know if you have any similar problems. Also note my OC settings below which are 100% stable according to Angry's OC guide comments on what should be considered 'stable' (see stability bench's in my first post on this thread). Note also that I have Cool and Quiet disabled and have not run into any CPU switching or stuttering in any of my gameplaying (BF2, HL2, etc.) I'll look into those windows hotfixes Ominous, thx.
  7. You should also check to make sure there are no EXTRA standoffs that are not directly under the mobos intended screw holes. This would cause a short also. I once had an Intel-only experienced installer try his hand at an AMD chip based mobo install and he didn't even check for unused standoff contact...sure enough it shorted out the board, but luckily not beyond repair (removal of standoff). I don't think the magnetic tip screwdriver would cause that type of malfunction but just avoid using it if at all possible. GL
  8. Thanks for the advice. I only have about 3mm of clearance between the top of the stock HSF and the Vid card PCB/AC 5 v3 cooler, but I'll check into that HSF you mentioned.
  9. I'd say unload that UPS for as much as you can get out of it (they have limited life anyway) and put it against the cost of a newer unit which will extend your UPS functional life anyhow. Consider it as just the cost of an upgraded UPS chassis with USB along with a new battery. To a person with older hardware, it could be like striking gold to finally find a com port UPS unit. Good luck.
  10. Angry > "This isn't meant to tell anyone that they stink and are not important customers anymore. " Just the thought of anyone saying this is hilarious (in a sort of gradeschool sort of way). Bring out your Dead !
  11. For what it's worth, I don't have an Expert but these same modules work perfect in my Infinity setup.
  12. The problem: when transferring data through and between my USB/Firewire ports for extended periods (throughput saturation for 5-10 minutes), the application or copying function will stop and my entire machine will lock up (data transfer stops of course). The odd thing is I've never encountered these lockups when OC testing my machine for the first 3 weeks. It's Prime95 stable at 15+ hours, passed OCCT, and SuperPI, all the video bench's and about 10 hours of sysoft burn in. Temps are normal (as when tested for OC stability) but the only thing I think it could be is that I noticed the chip that has that puny HSF behind my video card is EXTREMELY hot to the touch....could it be that this is affecting all my peripheral (external) I/O data transfers? If it is, I don't think I have any options because there is barely enough room for the chip's stock HSF. When I backup about 30gig or so between internal (SATA&PATA) drives (about 10-15 minutes), it's never locked up. Thanks for any thoughts you might have on this.
  13. Thanks for the encouragement EdLSmith. I looked for your entry but things seemed to have changed at the OCDB. After going over the new requirements that Angry has set for the OCDB, I'm going to pass on getting pics, posting pics, fixing the layout, re-testing with Everest etc. 90% of what he's looking for (sans pics and Everest) is in my post. I've spent adequate time and effort getting my results and analysis into this forum which was more out of an effort to give something back to the 'street' than making an o/c benchmark statement. Everything in my post is text so it's highly searchable . If Angry thinks this info should be in the DB and wants a screenie or two, I'll be more than happy to e-mail them to him. I can appreciate why he's doing this to the OCDB and know that pics are the proof but anyone doubting what I've posted.......well.......all I gotta say is o/c'n a PC is not really worth lying over. Thanks again,
  14. First of all let me thank the various contributors and ANGRY_GAMES for the excellent A64 overclocking guide. When I first read it my eye’s glossed over after the first 20 pages….I thought ‘how the hell am I going to assemble all these bits and pieces of knowledge into real practice and above all, feel that I was taking acceptable risks?’. I’ve printed loads of documents off the web similar to these first 20 pages but never felt comfortable enough to dive in and risk my cherry components . Needless to say I’ve never OC’d anything but a Vid card with coolbits, weak I know. So then I get to the “FUN PART” and there it is….a detailed walk-through for dummies…..I’m thinking…hell with what (little) I digested from all my overclocking reading….I can understand step by step what is being done here and WHY no less! Ok enough of the I’m-a-novice & kudos-to-Angry-and-friends. Finding MAX CPU FSB (not knowing what my max RAM FSB was going to be, I found max with a few multipliers so I had the settings if needed). Ended up at: (start with the meat, end with the salad for those who care) Max @ x11 > 224 fsb @ 1.525 vcore 2,464mhz Max @ x10 > 249 fsb @ 1.525 vcore 2,490mhz Max @ x 10.5 > 236 fsb @ 1.525 vcore 2,478mhz All at 40-41C idle. All tests stable (SuperPI/Prime-13hrs/OCCT/3Dmk01,03,05/AqMk3) I gotta say OCCT was the toughest to pass so it became my test of first resort!. Briefly how I got there. At x11 - no brainer. At x10 – Would not budge to 150 unless I applied a bios 1.575v (cpu-z reported 1.536v) and then temps jumped 4 C . At x10.5 – I’ve heard the warnings of half multipliers but frankly this was one of my top 2 final (stable) setting options! Oh and one FUNKY thing for the DFI folks – when I went one mhz higher to 237…after the bios and before windows…..it kept taking me to the RAID Array analysis DOS-like screen….and yes I checked….all bios raid settings were DISABLED…very wierd indeed ! Side notes. Undervolting below 1.475 would not work at all….I tried it since some of the best 4400+ oc’s I saw out there at the 11 multi, were in the 2600+ range @ 1.2v-1.3v….must have been with water or phase chg I’m guessing. All stable settings fell between 1.475 (@11mult only) and 1.525 (@10mult). I didn’t get one passing OCCT test at 1.550 until I pushed past to the 1.575v mentioned earlier (scary territory as I couldn’t find anyone else on the web with a 4400 vcore this high so I settled with 1.525 – besides my idle temps were in the 44C range). Finding MAX RAM FSB Ended up at: (there’s gotta be a drum roll for this, considering the effort it took) 240mhz @ 2.8v 2.5-4-3-6 (Tcl,Trcd,Trp,Tras) @ 1T 11,13 (Trc,Trfc) and Auto on all other settings except Async Latency 7ns & Preamble 5.5ns (more on reason for the auto settings later). Some of you may diss on this 'achievement' but it is my first OC attempt ok . All tests stable, same as above. Briefly how I got there. 1. I knew this RAM had low OC headroom so I had low expectations. As the guide suggested, I started with the std. ‘loose’ TCCD settings. 2. I started at 240, passed, 250 failed….bumped up Trp to 4 and then T1 to T2 and then had to work down to my final settings. 3. The ONLY thing that made me jump from 2.7v to 2.8v was passing the 2nd CPU test in 3DMk05…go figure. 4. I spent a LOT of time fighting with the optimal Trc and Trfc settings. I was stable with a Trp of 2 (at 240mhz), then I had to raise it to 3 (with no ill bench effects) and then finally got Trc to 2 over (Trp+Tras) at 11 and Trfc down to just 2 over that at 13. I swear I spent 3 hours fighting this (I’d held at 2 for the Trp for so long I somehow became ‘attached’ to it . 5. Ahh, why the ‘Auto’ settings you ask…, well for some unknown reason it would cause my recently-stable Trc and Trfc settings to fail. I tried numerous configs of these settings using information from the Guide but anything other than ‘Auto’ (tight or stable) resulted in OCCT test failure, so I just left them at Auto in the end. Criticisms or suggestions are welcome. Side Notes. Again, if it would pass OCCT, it would pass damn near everything else so I used it as my beginning litmus test for every new setting. Also, to help me visualize what my timing changes were doing, I took the cycle graph illustrated on the page before the ‘FUN PART’ of the Guide to see where some of my results were weak or excessively loose…this helped me logically fight to stay at a Tras 6 (with success I might add). Here were my major milestone PASS points in finding RAM stability for these modules. Others may find them useful if they have similar ram, mobo and bios. (Mhz,DRAMv,Tcl,Trcd,Trp,Tras.xT,Trc,Trfc) 225, 2.6, 2.5, 3, 2, 5, 1T, 12 ,24 240, 2.8, 2.5, 4, 3, 8, 1T, 12, 24 From here it was fight to find the tightest stable timings and maintain 240mhz. 240, 2.7, 2.5, 4, 2, 6, 1T, 12, 24 240, 2.8, 2.5, 4, 3, 6, 1T, 11, 13 FINAL In the interest of giving a complete account of determining my max FSB OC, here is the order of the RAM elements that I focused on from first to last (sometimes I had to go back and adjust earlier settings but here they are in general). 1-Mhz, 2-Tcl, 3-Trcd, 4-1T vs 2T, 5-Trp, 6-Tras and/or DRAMv, 7-Trc, 8-Trfc. Combining best CPU FSB with best RAM FSB !! I first drafted my top 3 OC scenarios with the following criteria. #1. The highest FSB possible regardless of CPU or RAM wasted (un-utilized mhz headroom). #2. The highest FSB that both A) yields the highest FSB AND B) utilizes the most CPU & RAM OC possible. #3. The highest FSB that yields the best RAM overclock possible. I doubted I had to resort to this though. Here were the setting options to obtain each objective above. #1. 10 multiplier x 249 cpu fsb @ 6:5 divider yields 2,490mhz with no wasted CPU speed and 33mhz wasted RAM speed. #2. 10.5 multiplier x 236 cpu fsb @ 1:1 divider yields 2,478mhz with only 12mhz wasted CPU speed and 4mhz wasted RAM speed. (all tests were stable with this multiplier) #3. 11 multiplier x 224 cpu fsb @ 1:1 divider yields 2,464mhz with 26mhz wasted CPU speed and 16mhz wasted RAM speed. I compared performance benchmarks between scenarios #1 and #2 (both prime stable) and #1 only outperformed #2 by a range of 0.2% to 1.1% through all of the tests. Although the margin was small, this further supported ANGRY_GAMES’ assertion that MHZ is king regardless of how you got there or how much of your precious (my precious) RAM or CPU overclock you had to give up to get there !! A solid system was achieved with the #1 option. This was a real heartbreaker after I spent all that time tweaking my RAM settings, only to end up using the RAM at 207mhz. On the bright side however, I got to take advantage of the super low latencies, not to mention I now know what the limits of my RAM are when I place them in a future PC build. Final Configuration and Benchmarks Final System Configuration: CPU ………………. Athlon X2 4400+ 1.525v @ 10 x 249 2,490mhz HT4(1,000) << A a bit faster than the X2 2800+ ! ...........HSF - Thermaltake Big Typhoon w/AC3 - Temp (Nv Monitor) Idle 25C Load 43C RAM ………………. OCZ PC3200 Platinum 2x1gb Rev 1.0 TCCD 2.70v @ 2Tcl - 3Tras - 2Trcd - 5Trp - 2Trrd – ..........10Trc - 12Trfc – Timing 1T - Preamble 5.5ns - Asynch Latency 7ns - Burst Length 4beats (all other settings Auto). HS – OCZ Heatspreader Vid Card ………….. Leadtek Nvidia 7800GTX 256mb o/c 497/1,360 ...........HSF - NV Silencer 5 rev 3 w/AC3 – Temp (NvMonitor) Idle 39C Load 58-60C Mobo ……………… DFI Infinity NF4 939 Rev A 3 – System Temp (NvMonitor) Idle 34C Load 42C (rest is in sig) Tests/Benchmarks: Sandra Burn ……… 19hrs. Prime95 Burn……..13hrs. Memtest86………… 16hrs. SuperPI 32MB……. 29m33.391s OCCT……………… Pass Sandra’05 CPU MIPS ….. 22,555 FLOPS ...10,330 RAM INT…….. 6,045 FLOAT… 6,001 AquaMark ’03 GFX……. 17,969 CPU……. 11,634 Total…….101,382 3DMark ’01……….. 31,925 3DMark ’03……….. 18,781 3DMark ’05……….. 8,839 Final thoughts. Always, always, always record every change made before each test along with all pre-existing settings (an excel table works). What you think may have been a severely failed test (early crash) could shed light on what to do next 10 or 15 iterations down the road. If this is your first overclocking project, and you’re going to do a nearly comprehensive OC analysis on your rig, plan on spending 15-20+ hrs on it (excluding ‘Prime’ or ‘burn in’ test hrs.). It took me about 25hrs before I was happy. I welcome any suggestions, comments on testing method or ideas for getting a higher OC. Thanks to everyone here at DFI-STREET !! Sincerely, YYZ
  15. divinemisfire, thanks for that detailed reply ! Right now I have the divider at 1:2 (ram fsb is only at 100) as recommended in Angry's top CPU FSB how-to doc. My temps at the stable 1.525v 224x11 were a maxed at a relatively modest 40 C but this is winter and ambient is at 23 C...but in the summer here in the attic it gets a bit hotter so I don't really want to get my CPU temps any warmer....the TT Big T is working wonderfully. Although stable, I must say during OCCT, my mobo chipset HSF was too hot to touch !! but like i said, it didn't fail....I can't put a better HSF on it because the NV Silencer takes up the whole space above the HSF! I won't worry about it much but it sure seems like an insufficient HSF for the amount of heat generation...hope mine doesn't fail like you said it might. I'll try the 10 multi when I try to merge my max cpu and ram fsb's. Until then I've got to find out my max RAM fsb.
  • Create New...