Jump to content


  • Content Count

  • Joined

  • Last visited

About Aeka

  • Rank
  1. Just a heads up, I stoped working on this board for the moment. I gotta work on my web page(the one that will end up hosting my mods ). And my stupid myspace page lol. And... I'm working with nf4 bios'es too, I gotta get the orignal's compiling 100% exact so I can start making mods for the: (I'm getting a dual core opty and a sli-dr this christmas) (It does'nt mean I stoped woking on this board, but when I do get the new one, I might not beable to use this one for a while, until I get some more fans and some more psu's) (Hands on experience on the nf4 may help on this board too ) nF4-D nF4 Ultra-D nF4 Ultra-D/G nF4 SLI-D nF4-SLI-DR I have'nt looked at the expert bios'es yet, I'll find one of the same ver to see if it's actually the same board or not. Later on that is. Anyways my testing method for this board, for both my data and my samsung. This will take a bit of explaining, so lemme pull out some examples as I can think of them. 11-3-2-3, 9-13 . That is one of the basics. It works even with crappy samsung, I mean really old tcb3(there is newer tcb3 out there though that does much better). For example of some old tcb3 samsung, I think, somehting like this anyways. 4-4-2-3, x-xx 0-0-3-3-0-6-3 It would look somehting like that on really old samsung. Actually, we'de need to up tREXT all the way up rpoabbly on old samsung so... 0-0-3-3-3-6-3 Good samsun even needs up'ing tREXT. Probably not the newer stuff, but stuff I had, some tcc5 I needed to at certain speeds. Not at 250, ot at 266, but at 275 I needed to. With the tcb3, I needed to at 250 straight up. With some newer tcb3 I had(just a tiny bit older then my tcc5, a 512m stick), I did'nt need to and it was running 266 np(I did'nt stability test, just in memtest it was stable so I did'nt care, I was playing and did'nt have time to play more). Anyways what config of the 2 is faster and or more stable? Well, at the lower speeds, the lower tRAS and higher alpha's are faster. But..., it does slow you downa bit, of course because of the alpha's. You gain more then you lose though . At higher speeds though, it's the oppiset, the lower alpha's are more inportant the the lower tras. I think that is wierd :, ohwell, the last time I beched it it came up like that so... : What is the most stable? The higher tRAS . So try this for starts: 11-3-2-3, 9-13 0-0-2-2-0-0-3 cas2.5, cmd2 (cmd2! ^^, more on this in a bit) This is the tighter of the slack tras setups. You can tighten it down more by using cas2, and... tRP comes next to tighten, then tRCD-R. (I will explain this in a min too). The slackeest you would want to use ^^ ? 11-4-2-3, 9-14 0-0-3-3-3-6-3 Or up to 11-14. The main timings up to 11-4-3-3. Lemme go into detail the mian timing order that I check. tRCD-W comes 1st, see if this goes lower tRP comes next tRCD-R comes after that. Then lower tRAS (depending of course on that one alpha timing). If that one alpha timing is 0, then it's still possible to use say 10 of tRAS potentially , maybe better . Then lower tRC, you shoudl beable to lower this all the way down to 9 straight up ^^. Then lower tRFC, which will depend on how high tRAS is in the end, and how high of speed you're running. The higher tRAS, and probaly the rest of the main timings, the higher tRFC may have to go. For example, at lower speeds you may beable to get away with 11 tRAS per say, and 13 tRFC. But at higher speeds, 14 tRFC . 10 of tRAS makes it easier to use 13 of tRFC btw. Another value for that one alpha to use perhaps, is 2. That can soemtimes help you use a lower tRAS()but it will not allow you to lower all the way down to the highest of the mains :, at least I doubt it will on anyone's system ). I know that was proabbly a bit tough to read, I dn. It's not like this is a sticky guide thingy . I did'nt plan this out lol. Now to cmd rates. This is what you'll probably need absolute. 0-0-2-3-0-x-4 (the 3 and the 4 are important in cmd1, for me anyways). I don't see anyone else using low alpha's so..., I'll just say for now it has to be this way at min ^^ This will aloow cmd basically. However, now comes your memory into play. On my mem, at 250+: 11-4-2-3, 9-13 (I think 13, maybe 14) 0-0-2-3-0-0-4 cas2.5, cmd1 Which sucks @ss :. It's actaully got it's up sides but really I dont' think it's worth a darn. However, with your mem, it's worth a shot, it's allways worth a shot to check up on . To see if you can do at least 3 timings with cas2.5 and cmd1. Here's my tightest timings: 3-3-2-2, 9-12 0-0-2-2-0-6-3 (6 because I got to tight with it, but I like this setup ) cas2.5, cmd2 (memory comes into play here..., I could possibly lower that one alpha tiny bit to 5 and some other things if I had better mem) My slackest(adata): 11-3-2-3, 9-14 0-0-2-3-0-0-3 cas3, cmd2 This got me 280 prime stable btw a while back. An old recored really. I got screwed over in the long run because of a psu isue (ocz). I since moved to the above tightest for everyday use, and I have gotten 275 stable again but it's hard at the moment to do much with that(cas2.5 timings though..., tight alpha's). My slackest with my samsung: 11-4-2-3, 9-14 0-0-2-3-3-0-3 cas3, cmd2 I gave this stick to my mom, max stable was 277x9, in dc in combinatin with my adata(yeah I mixed them). One thing I did'nt really explain is tREXT. As you can see above, I used 3 with my samsung, but not with my adata. In single chan mode, tREXT above 2 sucks. You can litterly lose 100's of megs of bandwith by using a value of 3. In dc mode however, the loss is neglagible(bad spelling I know) . So only go to tREXT of 3 if you abslutly have to. You shoudl only need to above 270 btw. If you have to use tREXT of 3 at a lower fsb, then it really means that you're memory sucks @ss. I'm sorry if that is the case, I hope it's not. Just because I said that, it does not mean 275 is not out of the reach . I got 278 memtest stable with super old samsung tcb3 before so... All is not lost if that is the case, you will NEED dc thuogh to keep any sort of bandwith going in that config. Again the max timings, seperate from each other's config(meaning could be diff alpha's, etc): 11-4-4-4, 11-14 (up to 16-18 for playing around) 0-0-3-3-3-6-4 cas3, cmd2 If you can't manage to get higher with those, then you're not gonan get higher fsb with even slacker timings. It is'nt gonna happen. One more thing before I finish this post off. tighter is better. Try to keep it halfway tight. You got room to play around, but note that the max potential is reached when you have the tightest timings that reach the max fsb. When you slack for no reason, you will lose stability, (though you can't learn if you can't slack I know). You can slack with my bios, just not with the optimal interface from 263+ - 274. You can slack at a slower fsb then 263 though with optimal cpu interface. But you can't after 263+ in optimal mode(aggressive is not effected by this). Once you reach past that area in optimal, you can slack, but only do so if you need. You can slack, but like I said, at those speeds, and everythign else, it's best to keep it tight as you can. So if you fund a setup that works at that speed, keep with that and make it tighter if possible(the tighter the better at those speeds). Remember that the last release so far was atiny bit wacked out. So if you test 262 and 263(maybe slightly higher and lower too), ensure you stay away from 8.5x and above. Once you hit 266, 9x is ok. 250 10x was ok so... Just a small quirk I can't fix without screwing over 275 for now. (The 728 bios kinda sucks I found out..., I had futher testing to do but I got a bit busy, for the time being) I'm sure there's other alpha timings out there to use. Those were just my kind of setups.
  2. You guys are right. It made snce that there was no 2xx 0r 8xx opty on the 939 so I looked it up and sure enough, the 265 was on the 940. Funny though, it's supported in the bios... So pretty much what I asked, if numa was supported, not really answerable. So no... The other ?, if 1 8xx opty was faster at the same speed as a single 2xx or 1xx cpu. Well, I dn..., not really the place to ask I suppose (especially since there is no 2xx or 8xx on the 939 apperently). AG/HG, look it up . Bah, screw it, I'll attach a pic just to be nice about it. Hmm can't attach anything... seems a prob with opera... I hope I can find a way to upload it before this thread gets closed :... http://www.voguehost.com/ims/u/ajdhsw/1/nF4%20Pro%202200.jpg That is a pic of a server board with the pro 2200. I also have a pic stored away of a desktop board using that same chip, not the same quality though(the above is my best quality one of that chip). As you can see, 1 more 0k resistor compared to the SLI chip. The Ultra chip is missing 1 0k resistor compared to the SLI chip (2 compared to the pro 2200). The nF4 (standered) chip is missing 2 more 0k's, 2 compared to the Ultra, and 3 compared to the SLI, 4 compared to the Pro 2200. The Pro 2050 is the asame exact thing. But it's got another config of resistors. It disables USB, and supposedly IDE and whatever. It was meant as a secondary chip to go along with the pro 2200 for sli 16x mode (40 total lanes). Most if not all boards using the 2 chip combo have each chip conencted to one cpu socket. Supposedly though they can doit via one socket, abiet a bit more limit perhaps I dn. I don't care lol... It's not like we could set that up on these boards. For 1, no one woudl want a 2050 as a single chip board anyways. So even thug I'd like to find shots of it someday for the heck of it, I woud'nt ever try making a board into one. (For example, I'd actually mod a server board to have 2x 2200 chips instead : , even though it would do nothing but change a device id). There's also a 2x setup for sli 16x mode, for desktops now I guess. It uses the SLI x16 chips, I'm unsure of the exact spelling thoughj since I have'nt looked around for the chips yet. I would guess they are the same, 2x of whatever chip, probably some desktop one. Perhaps another device id, I dn. Maybe even a who new chip ^^. Regardless... I felt as if I shoud'nt hold out on that. Even though it's nothing. This is a portion of what I saw in the bios: (03/29/06 I think) nForce4 Standard nForce4 Pro nForce4 Ultra nForce4 SLI AMD Processor Model Unknown AMD Engineering Sample AMD Engineering Sample XX AMD Athlon 64 Processor XX00+ AMD Athlon 64 X2 Dual Core Processor XX00+ Mobile AMD Athlon 64 Processor XX00+ AMD Turion 64 Mobile Technology ML-XX AMD Turion 64 Mobile Technology MT-XX AMD Opteron Processor 1XX AMD Opteron Processor 1XX HE AMD Opteron Processor 1XX EE AMD Opteron Processor 2XX AMD Opteron Processor 2XX HE AMD Opteron Processor 2XX EE AMD Opteron Processor 8XX AMD Opteron Processor 8XX HE AMD Opteron Processor 8XX EE Mobile AMD Athlon XP-M Processor XX00+ AMD Athlon XP Processor XX00+ Mobile AMD Sempron Processor XX00+ Mobile AMD Sempron Processor XX00+ AMD Sempron Processor XX00+ AMD Athlon 64 FX-XX Processor AMD Athlon 64 FX-XX Dual Core Processor Dual Core AMD Opteron Processor 1XX Dual Core AMD Opteron Processor 1XX HE Dual Core AMD Opteron Processor 1XX EE Dual Core AMD Opteron Processor 2XX Dual Core AMD Opteron Processor 2XX HE Dual Core AMD Opteron Processor 2XX EE Dual Core AMD Opteron Processor 8XX Dual Core AMD Opteron Processor 8XX HE Dual Core AMD Opteron Processor 8XX EE CPU0 Memory information: CPU1 Memory information: CPU2 Memory information: CPU3 Memory information: They seem like they're all enabled too, I checked the last entry of 8xx and the cpu# entries, those were enabled in the bios. I'm am pretty sure we'll never ever beable to use 4 cores on this board though lol. Still wierd. It's just something I noticed a little while ago. This is what I meant by the cpuid mod, if it would work(the nf4 pro mod would likely work though, this next one is up in the air). Replace this id: Dual Core AMD Opteron Processor 1XX With this one: Dual Core AMD Opteron Processor 8XX And see if it's speced as an 8xx cpu afterwards. It works on the socket 462 platform lol, so maybe it'll work on this platform, I dn :. It was something I was gonna do for fun. Just to show you guys what I was getting at as a side topic. The main topic being numa lol, and now I understand I'll have to mess around myself to figuer that out, if I can do anything :. Sorry about the small typos btw.
  3. You are the man I want to talk to btw lol... Yes I know this . I know these things are unsupported straight up. What I found just a min ago is that the 939, period, only has one mem controller. So no numa support possible. However, if it were to be actually 2x 64bit controller(which it could be I dn), then it's possible maybe. Then again that would yield little or no increase in bandwith. Comparing a dual core, in sc with numa a support. With a dual core, in dc with no numa support. Whatever I guess . Seems not important now. The chip mods I talked about. I got pics for the: nf4 nf4 ultra nf4 sli nf4 pro 2200 I miss pics for: Pro 2050 And sli x16 I woud'nt want to mod into those... I dn if the sli x16 is the same chip, seems like it though. The pro 2200 and 2050 is the same chip as the nf4, ultra and sli. Some brands of mobo's actually came with a 2200 on them. The 2050 disables stuff like the usb controller, and perhaps a few other things like ide. Regardless I was gonan try adding the 0k resistor for sli - pro 2200, since the bios does have support(not offically) for the nf4 pro in it (I looked). I wanted to ask you something though. I was pretty sure you had run an 8xx opty in one of your boards once(then again you might not have). Have you seen any diff at all in speed compared to the 1xx ver of the same type, abiet single core or dual core at the same speed? Apperently the 8xx ver's run a bit better then the 2xx ver's ona mult socket board with multi cpu's(edit: with 2 cpu's btw). But as a single cpu, I was a bit curious, if it possible enhanced speed just a tiny tiny bit or not. The reason I ask is because it seems possible to mod a 1xx or 2xx into the 8xx by modding the cpuid alone. I dn yet, but I will be trying it when I get the board and cpu this christmas, it won't hurt to try anyways, simple mod. If it does'nt help anything, and it does work out as a mod, then I guess it won't hurt anything like the chipset mod lol(both not helping at all in speed or functionality, just to be cool lol). And again, I totally understand opty's are unsupported and mods of any kind are unsupported . Btw, I found these in the bios as cpu id's supported (not iofficially btw). I just think it's wierd lol, you don't have nor do I expect you to respond on this one. Mobile AMD Athlon XP-M Processor XX00+ AMD Athlon XP Processor XX00+ (Edit: These are socket 462 cpu's built for the 939 socket ^^, odd...) (Edit: Actually now that I think about it, these might not be 462, they are probably reguler cpu's for thge 939, oui I'm dumb..) (Edit: :, oui they are 462's, the 939's are labled 64 on them) I found in the 01/25/05 bios for the nf4. Siince it's the 1st one I'm re-assembling.
  4. I was wondering about this for a while now... I did'nt know it actually existed though until a min ago(I did'nt know it was called numa). I can at least try to explain it to some of you that have never heard of it. You take 2 cpu's, each cpu has it's mem controller right? Well, each cpu can independantly access memory for a potential 2x bandwith increase. I was wondering, since there is no dual socket dfi board... If use a 2xx ver of a dual core opteron, if you could have numa, on a single socket on these dfi boards? I don't obviously have the 300x so on dollars to get one, I plan on getting one of those 165 dual cores for the moment. Regardless, if it is possible to use numa on these boards(sli-dr), then I'd be willing to pick one of those more expensive 2 way dual core opty's in the future . I would assume the config would have to use all 4 slots in order to use dc on both cores with numa support... And of course I have a supported os... It's just something I was thinking of, along with modding my chipset to the 2200 pro just for th heck of it lol(though it won't do a thing). I can't seem to find 2050 pics though, all I find are faked pics : (I wanna map out all the chips). I gotta find sli x16 chip shots too... : As a side note, I wonder what node interleaving is.... Anyone know off hand?
  5. A shim... I would'nt use one. I used to use on on the 1st gen xp's, but after a while I tossed it to the side. For one, they are'nt even close to being level to the core. Not that you'de want it higher then the core for sure, but close to level as possible... Shims though, mine at least, was probably half the height of the core. So in reality, I think it did'nt help a thing. Save your money to buy another tube of gel or something instead . That's what I'm saying anyways...
  6. Just a small update... Disabled "might" mean 4x. Meaning 64bytes in dc mode, or 32bytes in sc mode. There's something I thaught about before too. That enabled, may actually mean: 2x in dc mode. 4x in sc mode. I know it sounds complicated :. The best way to deal with this setting straight up is to leave it onauto, and hopefully it'll read off the spd. And not some bios table thingy :. If it reads off the spd in auto, most if not all ddr has 4x as a supported option. So it'll run in 4x. Some ddr (my tcc5 did'nt, but my adata does (by default) has 8x supported. The bigger the burst, the slightly more bandwith you'll get. However, samsung flips out at 1st on the larger ones (my adata fliped out on page mode at 1st). Just break it in with a few runs of 1m superpi at evbery 5-10mhz fsb until you get back to were you were(if need be). Then finilize the break in of this setting using 32m superpi at you're final fsb (if needed). I've done this before myself on my nf2, it's really not a big deal to break this setting in . You can break in all the way to the max np when it comes to this setting most likely. Say for example you can normally run 300mhz mem. But if it's on auto, you can't. Try stepping down to 250, boot that sucker up, run 1m pi a few times, probably 3). Then step up 5 or 10mhz at a time doing the same, until you reach 300mhz again, which you should beable to using that method. Then run 1m pi on that speed a few times, then jump to 32m. If it fails, try a few more times at 32m and if it's still failing, work your way up from 1m through to 32m (2m, 4m, 8m and 16m on the way). Once you pass 32m once, doit it again, then prime and s&m . Laters
  7. Take a gander at the attachment I'll post. I just started with this night. I do this when I'm like looking into buying something . (I'm buying a sli-dr and a dual core opty, hopefully : ). I got grabbed some of the settings out of the sli-dr/ultra-d bios, 03/29/06 I think it was. Using modbin. Later on I'll port it to asm and take a real look at it to see what there really is, etc etc. Anyways this text is incomplete. And really it was meant for me to look at and and play with, to think about lol. All in prep for the new deck, my main deck for a while until I'll beable to get this one that I'm on now going again (because I'm poor so I'll have to use parts form this one temp like) Tell me if it helps you out in any way though ^^. Before I finish this post off... Stupid attachment got renamed :... And personally, when it comes to that setting, I'm gonna use auto, because I'll be forcing "page" from the memory it's self(I mod my memory). I could use 1x, or 8x(not selectable in the bios), but I'm gonna use page. Edit: Mmm when I goto to rename those settings I should'nt use bytes in the setting name... That and rename Auto on that one setting to SPD... Blah blah blah
  8. I'm not sure.... This newest modbin(the one you have) does run and save files in windows(nt), but I don't know if I would realy on that one (but it does actually doit). I've heard the the older ones will too if you right click, set a compatibility mode thingy... But I would defnelty not realy on that one. I use dos personally. Win95b-c and winme dos(win2k3 dos actually, same thing allmost) work fine. No extras needed at all, no himem.sys or em386(or umbsys). I think me dos has himem built in though, it seems like it does lol. Anyways it's "supposed" (remmember I said this later ver is little quirky) run straight up in those, 98 I'm sure would be ok too. If you have dos support on your sata you're all good. The sil 3114/a controllers have dos support at least in ide mode, I dn about the raid modes. I have probs once in a while with all sorts of diff apps not runnign at all on my ntfs partitions, in windows. I dn why really. When I run into an app like that I throw it on one of my fat16 partitions and they run fine. I dn if modbin is such an app or not. Probably is, a seccy... Hmm it runs... A seccy... It won't run when I throw it a few dir's deep though. I'm sure fat32 would be fine if you did that. However, if the hd is small enough (just under 4gigs), I'd set it up as fat16, whatever floats your boat . Make sure it's something you can get access to if you're screwed over. I mean, after a bad flash, when recovering with a 2nd chip, ensure it's something you can access right away to re-flash the original chip asap. Also, allways re-open the bios with modbin after modding it. Unless you're absolutly sure abut it... Like if I used my tool to make a bios, I woudl check it a few times when wokring on the tool, to see if it was modbin compatible (seeing if I can make a 100% original is half the job). In the nromal ways though, cbrom and modbin, you shoudl allways try to open it up afterwards in modbin. To see if it'll load it up, if it does, just exit without saving. I've had modbin kill bios'es(probably actual instability back in the day) after moving menu's around to much :. Plus the fact that any modbin can be a bit quirky even if your stable(it epends on what you try to do or try to view even with it)(yes even viewing can screw it up sometimes so take steps). That's why I say to do that. Taking steps... Use modbin to take a look around at what you want to do, figuer it out and remember exactly what you want to do with it. Then exit it. Then re-open the bios, and do that, and only what you wanted to do, no pit stops . Then of course double check... Remmember sometimes you have to take a few steps at a time to track down oddities with modbin's mods. You can allways mod that modded bios more after it's verified. When flashing a bios.... No one seems to listen to this though :. awdflash bios.bin /py /sn /wb /cd /cc /cp /LD Keep a working bios on hand next to the modded bios too. So yo can get access to it, and quickly if you mess up and have to use a 2nd chip to get back. When flashing a dead chip, flash twice, sometimes you need to flash it 3 times it seems, but that only happened to me once, 2 times is good and it should allways be done on a dead chip. Three's a cmd string for awdflash to exit to doss after flashing but I dn off hand what it is. I just use F10 . Instead of F1 I think it was to reset. (Just hold he right arrow and it shoudl re-complete the previous cmd for yeah to flash again , if not, type it out again, or run the batch, whatever you do) Just a few more things . NEVER EVER enable bios flash protection and try to flash, it ALLWAYS kills the bios. If there's a such thing as a stupid bios option, that is it ^^. It does'nt even protect agaisn't corrupted bios'es or anythign, it just waits for ya to try to flash it lol, and it kills the bios :. Also remember to enable usb keyboard support if you have one, before you go into dos to hotflash(if needed to hotflash). It kinda sucks to get all there way there only to find out you can't do a thing :. That is if it's disabled by default. There is one more thing I can think of. I can't stress enough how handy a 2nd chip is(one that has a working bios on it of course). Even if you're not ocing, or modding. It can be usefull out of the blue with any mobo(as long as it's meant for that mobo of course lol). I moved a rig to my graparent's house way back now, when I got kicked out of my mom's. Worked fine over my mom's, got to my graparen'ts and for some reason, a corrupt bootblock. I know enough not to stick it by speakers and stuff like that. I did'nt mishandle it, it just died. And when a bootblook msg pops up on you ^^, you have a 50/50 chance to recover from a floppy(and only a floppy). That is if the bootblock is even working, and like I said, half the time it is'nt(it's not the floppy you have to worry about lol). Luckly I had a board with a dual bios socket setup, and I had a 2nd chip allready in it all ready since I modded the thing once in a while. Do yourself a favor . Get a 2nd chip. It's one thing to wait for that 2nd chip, but waiting for that 2nd chip when your rig is down, not even knowing if it's gonna wok out, really sucks :. Trust me on that one, even if it's only a week or whatever, get that 2nd chip before you kill the bios (not saying you will at all, but you really should get one, even if you're not modding, this is defently the type of board that you want one with(not to say anything bad with dfi, peopel's...)). You get that 2nd chip and it's finky :, that sucks (happened to me, it really sucked, took me hours to get it to boot). So get the 2nd chip, flash it twice, and verify it . I believe Tmod is the man you want to talk to about bios chips . (See if he's got sst lol, I dn if he does or not, sst are good ones, they have speed grades too btw(so does pmc)...., my sst is a 33mhz chip, it's pretty quick, there's faster one then this too, I don't know what he's got though and anything is ok for a 2nd chip even if it is winbond or pmc)
  9. That ver of modbin is a pos, but it is the latest on the net for the public tweakers . It does work, but like you've noticed, it's very bugg in the menu's and it won't save a file sometimes. The trick is to remove the temp files before you run it , from previous runs. Also saving as a diff name might work better in that ver I'm not sure off hand. Like I said though, it's menu modding feature's are a bit wacked, sometimes working out somewhat, other times totally corruptiong the bios, or apperently, from what you say, not doing anything. There is another way, to doit all by hand. That's not as easy though. I can rename stuff. Repack, etc all by hand. Not using cbrom or mobind, even fix the checksums in the lha's, and the bios'es bbss. There's also the lha original file name and date... Name length, etc etc. What I have'nt gotten around to is maping how changing defaults. I've done it by hand once some time ago, I've just been to lazy to do any real work in asm for it in my tool thingy. Anyways to make this short. I'd reccomend you do the cmos save thing for now. For a medium, everyday setup (the highest, stable at all times settings no matter what sort thing lol). So you can at least jump to some halfway useable settings (I know what you mean, the defaults are allways lame in any bios). The defaults are determained by of course, if they're hardset as a default. If they are auto by default, they read off a table of settings. There could be multi tables of settings, each for diff bins of memory (baised off tCK straight up, the lower this is, the lower youre timings will be until they hit the min supported timings in the spd). Usually the bios doe'snt totally follow the spd. Lol, actually, it can decide to outright ignore the settings(because it was not programmed to even look them up in the 1st place.., not good, but ohwell). DFI boards do some ignoring too... Btw, before I take off here... Modbin, even if it's working 100%, it allways corrupts the living heck out of the bios. It'll flash for sure, if modbin worked right. But like I said, it's never the same again, never. Still, it's not a totally bad thing to do (as long as you got a mobin ver that is working good for your bios). The last "good" ver of modbin was: 1.00.38. It won't work on your board I'm betting... Any ver above this is way way buggy(totally not useable). That last ver, the one you're using now is halfway useable though, but barely.
  10. I use superpi for breaking in my mem. Memtest can work, test 5 looped. As long as you're allways passing it. Running that for about 30mins can sometimes get a stubbern s&m to pass. Superpi is the way to go though, it's a shame however that that it cannot automatically loop and or retry it's self (kinda sh!tty really). Just remmeber that once pi fails, you have totally restart the program. Otherwise it just sits there stupid like saying you failed every time you try it.
  11. There wer 1.35v mobiles too btw . I don't think there was a 2600 ver of that though, I think it was a 2400 of it max (2.4-2.6 were the ones to get back in the day). I think, and don't quote me. That there's some 3.800+ mobiles out there as desktop replacement cpu's. I guess they clock ok but I have no idea, I only heard that on this forum somewhere. Btw, our mobiles default at 133, not 100 . You can gt both 100 or 133 after a cmos clear though, it's the board being wacked for some odd reason (not reading off the right id or something). Maybe there's pins missing "in" the socket like many other boards, I have no clue :. Does'nt matter anyways. The default multi for the 2.600+ mobile is 15x I think. (don't quote me on that one) Which is essentually 12.5x... SIP wise it is, same timing anyways .
  12. Other HKR,, NV_Modes, %REG_SZ_APPEND%,"1120x630=1FE7;" ;[16x9 - 70 DPI] 240 200 170 150 144 140 120 100 72 70 60 I use 104hz. Why you ask ?? Messing with horizontal pixals in html is a nono.... You should only play with vertical pixals. Set horizontal by %... If that's why you had this poll anyways...
  13. You get alot more "ranked like" bandwith in memtest with the more ram you have. Other things that determain bandwith are things in the spd, how the cpu is broke in too, that's another. Most things are'nt measureable in any bench though. For example, lowering the tCK timing all the way down to 01h. Not measureable really in any bench at all, not even superpi moves much(it does but very very very little). Yet s&m's mem test doubles in speed, go figuer :. Heck I've gotten - latency in everest before, that's what hpapnes when you realy on generic benchs. I've gotten 0 as well. Totally unrealiable bench, but so is sandra :. Most of the time they do ok though, sandra however is one I don't trust because it varies to much on other's systems(on a p4 I saw it was way to low, were everest was about normal for that mem speed). The feal. How much faster is this or that, or snappier. Then the old benchs. They all play a roll. Timings are important in the sence, of what you want. Can you deal with slow sh!t... ? If you clock high enough, even the most slack of timings can be dealt with lol. However, there is nothing like tighter timings. You can outperform memory running at 250 with memory running at 180 with tighter timings. By a good amount too. Only in certain areas though. Measureable in fps even. But the diff the 250 has in those bench's, blows away that 180. Like it was never there, yet, the 180 was winning out by 5fps in mame's psx emulation ? ... Hmmm Voltages can determan performance too :.
  14. When I looked into this stuff along time ago I found: 2.800+, normally clocked to 3.200+ lv's easy I guess. Was cheaper then the 3.200+. Supposedly on avg it clocked better then th e3.200+. The mobiles. On avg, pretty much garrentied(bad spelling), a 2.4ghz oc . Or more... They were cheaper the the desktop cpu's ^^. At the the time that is. They were unlocked by default, the mobiles. Pretty much the only real way to go, $100 max and you got the cpu you needed at the time. You got your 2.4ghz or more easy with them. They were cheap enough for anyone to get. They were unlocked so going for the max fsb was just straight up. Thats' why . They clock easier then most cpu's. Excluding the durons though, those clock much easier it seems. 3ghz is doable on crappy(and I mean crappy after market generic coolers) :. The semprons have all 512k of l2, but they don't have it enabled by default. They all seem to clock on avg around 2.3ghz on crappy boards. The durons clocked to 3ghz on crappy boards... : Well one duron, most clock around 2.4ghz I guess. The semprons are defenlty unlockable. I dn if it's the same way everyone else unlocks them though, bridges and stuff, I don't really pay attention to that lol. I think they are cheap too. I don't really see a better core build out there then the mobile xp's though, in barton or thorten baised cores. Since they have all 512k of cache straight up, only the thorten mobile's have 512k though, there desktop ver's have 256k. All barton's have 512k. Anyways I hope that helps :. Mobiles are hard to find though. It's easier to find a 939 cpu these days. Then again people are still selling off there old decks ^^.
  15. I wanna expand on what I said above. http://www.2104.dk/forum/index.php?topic=362.45 Please read this. I screwed around with my ff and left it be for a while, posted about 2.0 there. Then tried to re-produce my previous issue with it. I could'nt :. I have a screenshot of firefox after closing, where it did'nt close, and at had been in my taskmgr 8 times (which before I had closed... all of them, 1 instance created 8 of them). It's not viewable if you're not logged in there, and I woud'nt actually want you to login there unless you were into that stuff, emu stuff mainly there . And apperently attachments are'nt working here at all... How nice lol. Hmm... http://www.savefile.com/files/201127 There Now can we vote on how lame I am lol ?, joking about the vote part but still... What do you think of the pic anyways ? Seriously... Maybe read the other forum's 1 page where I talked about 2.0.... Long though, very long. These are my tweaks to mozilla so far(the ones I've listed in a text anyways) Take any you see fit to use, if any. browser.blink_allowed false browser.cache.disk.capacity 0 browser.cache.disk.enable false browser.cache.memory.enable true browser.cache.memory.capacity 65536 browser.chrome.toolbar_style 0 browser.display.background_color #004066 browser.download.progressDnldDialog.keepAlive false browser.enable_automatic_image_resizing false browser.download.manager.closeWhenDone true browser.download.manager.retention browser.related.enabled false browser.related.autoload 2 browser.related.provider browser.turbo.enabled true browser.underline_anchors false config.trim_on_minimize false dom.allow_scripts_to_close_windows false dom.disable_window_open_feature.close true dom.disable_window_open_feature.directories true dom.disable_window_open_feature.location true dom.disable_window_open_feature.menubar true dom.disable_window_open_feature.minimizable true dom.disable_window_open_feature.personalbar true dom.disable_window_open_feature.resizable true dom.disable_window_open_feature.scrollbars true dom.disable_window_open_feature.status true dom.disable_window_open_feature.titlebar true dom.disable_window_open_feature.toolbar true dom.disable_window_open_feature.status true dom.disable_window_open_feature.titlebar true dom.disable_window_open_feature.toolbar true dom.disable_window_status_change true dom.event.contextmenu.enabled false general.smoothScroll true layout.enable_japanese_specific_transform true layout.frames.force_resizability true network.autodial-helper.enabled false network.dns.disableIPv6 false network.http.accept-encoding x-deflate,deflate,x-gzip,gzip,x-compress,compress network.http.keep-alive true network.http.keep-alive.timeout 0 network.http.proxy.keep-alive true network.http.request.max-start-delay 0 network.prefetch-next false nglayout.initialpaint.delay 0 prefs.converted-to-utf8 true update_notifications.enabled false
  • Create New...