Tamper

Olham, I hope it works out for you...unless you need a router, then a switch or hub are both probably cheaper. Router is only really necessary if you want to share Internet among PC's on your LAN. The cable should work assuming it's a proper crossover cable - technically, the machine(s) don't know how they're physically connected - they just know whether there's a server present that assigns IP addresses (called a DHCP server) or not. If not, both machines should check for other machines and then autoconfigure an IP address. If an Ethernet adapter is present (even with no connection) the machine will give itself an IP. Just to hazard a guess, I'm going to say the two machines were in different ranges. Go Start, Run, type 'cmd' and then (in command window) type ipconfig. This will tell you the IP address for the machine (again, this for XP, Vista would be similar). They need to be in the same IP 'range' (ex. 169.254.nnn.nnn typical) and 'subnet' (255.255.0.0). Regretfully, I don't know a lot about Vista - but I gather this same mechansim works as well - you can find out tons of stuff on autoconfiguration of IP addresses (called 'APIPA') on lots of websites (google to start). More info http://support.microsoft.com/kb/220874 HTH

As far as the setup goes: You need what's called a 'crossover cable' or a Ethernet hub/switch/router. The cable is by far the least expensive way to connect two (but only two) machines. It is a regular piece of Cat5 cable, that has the 2 needed pairs of wires inverted. I can give explicit details if you like, but if you don't know how to make the cable and don't have the connectors/crimping tool, you're better off just buying one. If you have a hub/switch/router, they simply both need to be plugged into the device with a regular Ethernet "patch cable". Both machines, when they boot, will use IP Autoconfiguration to assign themselves an IP address. (If you're using a router, then the router will assign each machine an IP). They'll both be in the same range/subnet, and shouldn't require any changes. Make sure both machines are in the same 'workgroup' (right-click MyComputer, go to Computer name tab). If not explicitly changed already, both will probably be in the same workgroup. Also, make sure that Windows Firewall isn't blocking the connection when you go to run MP (it will try if it's turned on). Best way to make sure you're connected is to share a folder on each machine; go to My Network Places on each and you should 'see' the other. That's about it, really,as far as the LAN part goes. (EDIT - the foregoing applies to XP - can't speak for Vista, regretfully - but should be similar).

Fascinating. I appreciate your telling me about it. My board is actually a 780iFTW - I wonder if the same applies to it as the 780i boards? I actually trained on electronics in the Navy, quite a lot - went to their best school in fact, but that was a long time ago and I am usually victim of being lazy...if I have to start doing all that math again, screw it, I'll just buy something different *lol* I do know that the newer boards make a big deal about using solid caps and other design improvements - which actually makes sense, based on the formal training I've had. People ("enthusiasts", I think we're called) do know a lot about this stuff because we're constantly jacking things up faster and faster, expecting more and more performance. Drives the market harder to make better products. *lmao* Years ago, no one thought much about the caps. The 1333 thing - you know, I had an Asus board based on the 680i. That thing was squirrely as all hell about memory beyond 800MHz, and though it handled the C2D 8400 (1333FSB) well enough, I only ever got the CPU up to 3.6G (1600). But the 780iFTW board, same exact physical CPU I can get to 1700 (3.83G) without a lot of heroics. To be honest, I just lost patience with all the effort; I got it to 27.5% overclocked and stable, and said enough was enough (for now *lol*)

"Depending upon the timings being "loose" or "tighter", you may have to nudge the voltage in order to stablize tighter settings. " It is often necessary to adjust the voltages when overclocking, but I just want to point out here that (what I believe) we're working towards is actually more akin to 'underclocking' (I'm not even sure the term exists *lol*). Joker seems - and many like him often are - concerned about overclocking; they'd rather not take any risks. And it's a perfectly well-taken point; you and I might not mind the risks (whatever they may be) but it's not for everyone. I am hesitant to treat this case specifically as if we're overclocking, because it seems wiser to avoid the impression that's anything like what we're doing. I prefer to say we're "using the same methodology" but for different results - opposite results, in fact. It's more about people's comfort level, really. To that end, I don't think voltage settings will enter into it. In my own experience the voltage settings are normally only necessary when you're "committed" to overclocking, and even then the heat issue starts to become more of a concern. Get started explaining all this to people who aren't comfortable, they tend to freak and just say "Forget the whole thing" *lol* If we're being totally honest, I think that may have had to do with Joker's initial hesitance to get into BIOS and memory settings to begin with. Of the three things I mentioned that apply to memory settings, I try to recommend that voltage(s) should be the last thing you change - and only then if you're committed to it, and you understand the consequences. Still, in this case, I believe it may only involve very small changes to the timings - there are at least two experiences related above that are similar - just enough to stabilize things. I'm thinking it may take only one step off what his current timings are to get it to behave (like what HPW related). As an aside, I'm interested in what you mentioned about the eVGA memory controllers on the 680/780 chipsets...can you share whatever references you've seen? I'd like to read up on that more, myself :yes:

On the subject of timings - The slower memory speeds (in Joker's case, PC2-6400 or 800MHz; in HerrPW's case DDR1) generally will run stable at tighter memory timings, as I understand it. In fact, it seems I've read that DDR2 memory's higher speed is offset (somewhat) in higher latencies. To 'compete' with DDR, given its higher latencies, DDR2 actually must run at a higher speed, else it wouldn't be "faster" (performance-wise). When DDR2 was still first being developed, it didn't compete well with DDR, because the latencies were still high enough that even doubling the speed didn't make up for it. As the chips improved, the latencies came down, and - combined with the higher clocks speeds - overall performance improved. Kinda confusing, sort of (but not really), but that's what I understand, anyway. For example: HPW's (DDR1) supports timings at 8-4-4-1 Joker's PC2-6400 (DDR2) timings are '4-5-4-11' (only slightly tighter than HPW's, and that, I suspect, because it's a good deal newer). My PC2-8500 timings are rated 5-5-5-15 - compared to Joker's, more 'relaxed', even though my memory runs at a faster clock rate. And I'm actually running at 5-6-6-21, after relaxing mine during my testing. The settings I'm currently using are above in the images I posted, as are the "SPD" for both JDEC and EPP timings. Joker, if you can get/run CPU-Z (www.cpuid.com) it will report all these timings. Easier than BIOS screens but to be thorough we should still look at other BIOS settings as well. Finally, I wanted to mention that there's not really a 'right' answer in this case. The rated speeds don't always work, sometimes people get away with increasing speeds by relaxing timings, and whether/how much relaxing helps still seems largely a matter of individual components and good old trial-and-error *uggh* As above, what matters is stability.

Also, does he have the latest BIOS for his motherboard? MB manufacturer's sometimes release updated BIOS for their motherboards that fix odd problems. He should check eVGA's website. Agreed, and an excellent point. I am not completely sure he is using an eVGA motherboard, though - just that he's mentioned a 780i chipset. Still, BIOS updates can fix issues like this, depending. It's at least worth checking to see what's there - whether to use updates is worth some consideration. The rule a long time ago was if there's an update, apply it. Nowadays the advice tends toward only doing updates if you have a specific problem (me - I usually try 'em anyway). I agree about the memory timings. My memory (DDR 1) is rated to run at 8-4-4-1, and it works fine for most programs, except OFF. For OFF, I have to leave my memory at 8-4-4-2. You don't say? Nice to know that I'm not losing my mind *lol* Seriously, it is very interesting that you've seen this sort of thing as well. Taken together your experience and mine seem to corroborate one another. I suggested to Joker that he set his BIOS to default, but he never indicated if he did or not. I don't recall seeing he did, no. As I said earlier, I think some idea(s) just got lost in the deluge. And I can certainly appreciate being wary of fiddling with things you don't understand - that's a good habit to be in! But, the trick is exactly as you said - record things first! (And never changing more than one at a time, as well - although I cannot truthfully say I never break that rule *lol*) Thanks for mentioning this - seems like many here want to see this figured out :yes:

Sure, I'll try: One thing - just to confirm, my board is an eVGA 780i FTW, and yours? (Probably only slight differences in BIOS, with the same 780i chipset, but thought I'd ask). What I'd be most curious about is the memory settings. Generally, this accounts for three things: - Voltages (this would include those for other system components, such as the "NorthBridge", or - to Nvidia - the "SPP" (website lists your memory as 2.1-2.2V) - Front Side Bus (FSB) speed setting, in MHz; yours should be 800. - The memory 'timings'; generally these numbers are included with the memory (website lists yours as '4-5-4-11' - these are pretty tight timings, actually) Most often in my experience whatever 'tweaking' can be done will involve one/all of these three. And, just so you know, we're not talking about doing anything for the sake of increasing performance (i.e. 'overclocking'). Adjusting these BIOS settings isn't necessarily always about overclocking. What we're interested in here is sort of the opposite of overclocking; it's about stability. Oddly enough, this is always a part of overclocking, because fast isn't worth a darn if it's unstable. Note that these things usually aren't necessary unless a system is overclocked, but never say always. I'm listing some things here because a. we don't as yet know what your BIOS settings are; b. something may be set by default that just won't work. Some examples of things that often increase stability (in no specific order): - increasing (in tiny increments) the voltages. Usually reserved for pretty jacked up overclocking; the system's components require more voltage to run at higher speeds. Bad side effect is heat. - FSB settings. Easiest way to fiddle around with settings; not likely to cause many problems unless set to extremes, but you need to be familiar with the range(s) of the applicable extremes. - Timings (sometimes called 'latency') Tons of theory about this on the web. Basically, lower latency = faster memory. But, in terms of overall system performance, we're talking pretty small differences. Usually 'relaxed' as other settings are increased for overclocking. "Relaxing" is to slow down the timings; 'tight' is the opposite of 'relaxed'. As above, your memory boasts some pretty tight timings, I'm wondering if that might not be a good place to start - but, without knowing more about your current settings, it's very speculative. Not usually good to guess a lot. First thing I'd do is look at what's already set up, take the time to note everything as is, and then look for anything obviously out of whack. Is it possible you could post BIOS screens here? If nothing seems obviously wrong, then you can try slowing things down a bit to see if it helps. Honestly, the most frustrating part of this is the absolute 'trial and error' nature of it...and there's really no way around it. Some prefer to go at a goal in small increments, stopping when they hit snags; others prefer to set things to extremes and then back off that until things are stable. (I know this sounds a lot like overclocking, but it's not; essentially you're using the same methodology as in overclocking but with different results as a goal). And, just for the purposes of your testing, I'd take 2 modules out and only leave 2 in. (In the simplest of terms, this is about reducing variables).

To be honest, I don't know what the root cause is. I'm still working on it - though, truthfully, once I got things situated where I could actually play a game *lol* I wasn't terribly anxious to start with the 'joys of trial and error' again. As you might imagine, it's frustrating to deal with a problem like this. Even though I do enjoy working on PCs, I also enjoy being able to use the things :) What I can tell you is that my personal machine is a 780i MB, which supports 1066 memory, and I use 2x 1G Corsair Dominator PC2-8500 sticks. Upon setting it up, as always, the first few hours were testing: SiSoft Sandra, Prime95, CPU-Z, CoreTemp, et al, just to see that everything is in order. Now, admittedly, I do overclock - but before anyone freaks out, realize I always start at defaults and return to them when there's trouble. Read on. Anyway, at the end of the whole utilities exercise, everything seems happy - so on to the games. CFS3/OFF runs without incident, check. Another graphically intensive game I run is Age of Empires 3. Now, keep in mind, this setup has already passed the utilities run, and shows no sign of instability at this point. When I ran AoE3, crash, crash, crash. (A lot like Joker, huh?). Adjust this and that over the next few hours; finally figure out that my memory - at the rated FSB speed - causes this crash. Even once I put everything back to defaults, so there's NO overclocking, CPU or memory. Eventually, I've settled on 930 FSB (which is technically 'underclocked' on this board and with this memory). As I said above, I have left things there for now, because a. I got kind of tired of screwing with it for now - nothing I could think of helped, and b. I just wanted to use the thing for a change :) So, there's one example. Also, on this machine, prior to changing the motherboard, I had a similar situation, with similar components - and also eventually wound up "underclocking" the memory back to around 935. The main difference(s) were different motherboard and memory, and that it was OFF at the time that I couldn't get to run (same exact symptoms - again, a lot like Joker, huh?) So, that's two examples. It's worth noting that, in both cases, the 'freezing/crashing/lockup' showed symptoms such as graphics corruption and/or sound 'sticking' (frozen/stuttering) such that I associated it with a heat, bad video card(s), sound card, or a driver. Nope, none of the above. All of them are fine, if I just keep the memory turned down. I've also relaxed the timings a bit, just as part of troubleshooting. (Again, I need to do some more work on it, because I could probably zero it in more, but it's definitely and directly a product of memory settings). Far as speculation...well, it is just that at this point. What I found interesting is that, in the BIOS, setting 'auto' does NOT set this memory for 1066 - fact, neither motherboard's BIOS did, for either of two different brands PC2-8500 memory. It always wanted to set default at 800. And, after studying this a bit, I found out about the JDEC vs. EPP 'ratings' on the PC2-8500 memory. In my case, even though the EPP for memory is 533 (1066), the JDEC rating still shows as 400 (800) FSB. I attached some pics here to show this. So, am I 'overcloking' 800MHz memory, or 'underclocking' 1066 memory? Apparently, it depends (*ughh*) I haven't really finished trying to determine why this is, or how to (maybe) overcome it - but, my speculation is that the memory is basically 800 that is rated to run at 1066. But, since the motherboard manufacturers don't make the memory - and memory manufacturers don't make the motherboards, it seems they'll all say "It should work", but they won't accept responsibility if it doesn't. I've read a bit, but I could easily be missing one simple, crucial piece of info I just haven't come across yet. Regardless, like I said - once I got it stable, I've left it, at least for now. Edit: Another reason I left it as is, is that after repeating the benchmarks, there's almost no difference in running it at 1066 - where one game will crash, and at 930, where the same game seems happy These experiences are what leads me to believe there may be a similar issue with Joker's setup. Even if not, it's remarkably similar in behavior. I welcome any thoughts/comments.

ALT-TAB works for me, but I noticed I have to use ALT-TAB to get out of OFF and to get back in (can't just click OFF on the task bar). In fact, I used it just this morning to grab and edit some screen shots while still running OFF.

- deleting "uisel" file: I read the thread linked about this, I don't think it's your issue. Biggest reason is that the thread describes a crash that basically occurs at a time totally different from what you see. Also, the uisel file has to do with User Interface selections (hence, 'ui sel')...your crash isn't symptomatic of User Interface Selections (menus, etc.) - it happens while running the game. - possible bad disk: Maybe, but then - if I recall, you were able to play CFS3 fine before, just that OFF would crash. Now - after some change (don't recall, I think un/re-installing) you can't even play CFS3. Still, you were able to before. So, is the 'bad' disk actually changing the way it's bad? I sort of doubt it. Even if you want to rule out a bad disk, arrange to borrow one, and use your own license/PID at install; nothing wrong with that and you don't need to buy another anything. (I'll even loan you mine, if you want). I urge you to reconsider the RAM. None of the things you mentioned (showing up fine, being identical, defaults set in BIOS, not overclocked) will keep you from having the issues I've seen on two different motherboards, each with the very same conditions (showing up fine, etc.). You mentioned that the bad disk and the bad uisel are the only two things you haven't tried. Actually, I don't remember that you tried anything with RAM other than verify it's set for defaults and not set for SLI. (You did, I believe, buy new RAM - but even that won't get around the problems I've seen). I don't recall reading much about how the memory is set (BIOS) and I don't believe you mentioned trying alternate settings. Also, as I said, you're using 4 sticks - I don't recall your saying you tried it with only two (not one, mind you, but two, in the proper slots). Not to be argumentative here, only trying to help. But I just don't think you've exhausted the biggest candidate. I run tech support for a group of about 150 techs and 90% of the problems they have trouble solving is because they overlooked something basic or assumed something - even professionals make this mistake. The very symptoms you describe, I've seen at least twice, on two different motherboards/chipsets, in two different games, and both times it came down to memory. Memory that, like yours, worked fine for any other game, and was otherwise fine for all accounts (benchmarks, Prime95 stress test, etc). Not at all saying the memory is bad - just that the settings can be 'finicky'. Besides, honestly - what do you have to lose? Write the current settings down (or take digital pics) and put them back if it doesn't help. Seems a relatively small investment of time/effort when you consider everything you've already put into it here. Whatever you decide to do, best of luck. I really would like to hear someday that you got this going - and what you did to get there. The enjoyment you'll get from OFF will be worth it.

Still going to wager any amount that, ultimately, this has to do with memory - not the modules themselves (they were swapped, as I recall). Rather, the BIOS settings that have to do with timing(s) and FSB settings, etc. for memory. Why do I believe this? Well, because I've experienced precisely the same problem - with SLI/without, after swapping not 1 but 2 sets of (different brands) memory, different video cards, 2 power supplies - the whole gamut of stuff that's been tried/recommended here. I keep lots of parts and build PC's, so having parts to swap around helps. But no matter what I threw at it, nothing seemed to do the trick. Yes, it behaved like heat somewhat (happened after a period of time), and only happening in OFF while other games worked fine (although it had also occurred at other times, each time in only one game). Someone suggested setting BIOS to default; might not be a bad idea to try that (if you didn't; I admittedly have skipped a lot in this thread). If you take the time to record settings before changing anything, worse case you can put it all back and you're no worse off. Focus on the memory settings, keeping in mind that what a machine detects 'automatically' may or may not work. These things were all mentioned at points, but then got displaced by other ideas - so I'm thinking none of them were ever actually tried. Something else I noticed - you have 4 sticks of memory - I know you changed memory, don't recall if you had 4 sticks previously. As a test, you should really try only two at a time, and make darn sure they are placed in the two same-color slots, starting at the slot closest to the CPU. (Running memory in 4 slots and/or in the colored pair that starts second from the CPU has definitely caused the same sort of corruption/lock-ups you report). On my 780i board - same as yours, I think, it fails to boot if I put the 2 sticks I have in the slots farthest from the CPU. When I used 4 modules (same ratings) I experienced the little funny-colored squares around certain screen elements, graphics corruption-kind of thing you've seen. I thought heat and video cards, too - nope! It was a memory problem. Back to basics; took 2 sticks (of 4) out, put the BIOS settings back to lowest (800MHz, even though memory itself was rated PC2-8500 or 1066MHz). It worked. Understand that, even with PC2-8500 memory, the JDEC SPD (not short for 'speed' like most think, it's actually 'serial presence detect') is ONLY 800MHz (PC2-6400); the higher settings are 'EPP' (Enhanced Performance Profile or some such) for memory that is supposed to be certified to run faster. Some BIOS will automatically set to the EPP settings, and even though the memory's in perfect working order and rated to run at 1066 FSB, it just won't. And I learned this while running OFF (or, more accurately, watching OFF crash a lot *lol*) Finally set memory FSB back down to 800, non-SLI enabled, and it works like a charm. (I've since learned I can bump it up to around 935 without trouble - but it won't run at the rated 1066 FSB for anything).

Sort of followed this from other forum. Still thinking memory issue (even though every thing else works OK - yes, I've seen this, first-hand). Have you tried different BIOS settings (for example, "relaxing" the memory timings)? And, even though I think you said you tried different memory, if it's set up in BIOS to "automatically" configure timings, etc., can still be a problem - the 'new' memory may be failing for the same automatic timing settings. Sorry it's such a buggar, but - like the Service Manager who hired me for my first tech job after the Navy - "Sometimes, you just gotta out-live the problem". Don't give up :)