Redshift
12-06-2004, 11:16 AM
Heh. I've been lurking for a while, and I genuinely never thought I’d be making one of these sorts of posts, but here goes:
I’ve been without a decent computer and gaming platform ever since I sold my old 2.4B Northwood @ 3.3 to work, which was about 6 or so months ago. The intention then had been to wait for the FX-55 (which at that point I believed would be a San Diego cored 90nm CPU), get a Mach 2 GT and either do an SLI setup or whatever top-end card ATI had to offer at the time. It was a good plan… only reality had to complicate it! Damn. I find out that the FX-55 was to be based on a 130nm Sledgehammer part without SSE3, and that nVentiv went out of business; building my own phase change unit would be cheaper and a cool project, but I simply lack the time and tools. Furthermore, I didn’t really want a Vapochill LS, so I seriously began to reconsider my plans. I’d always wanted a dualie, not because I ever thought it would be better at single threaded games but because I could multi-task a lot better and could actually do productive things on it (sure a little gaming on the side never hurt anyone). Then, reality has to go and complicate things again… turns out the FX-55 clocks really well, and Mach 2 GTs are available again. So now, I’m faced with a terrible choice: abandon OCing (I’ve got no illusions about trying to OC Opterons on a Tyan board) in favor of SMP, or build essentially the original system I had planned, but with a very fast processor that will be less future proof. Let me explain the last comment: I think that with the advent of mainstream dual core CPUs application developers are going to have to start making multi-threaded apps and games. SSE3 and the other optimizations included in the rev E. core are also important to me. I know that the performance impact probably won’t be as significant as other factors, but if I am going to plop down as much money as this is going to cost, I want to be happy with it afterwards – and not feel like I missed out on something better by being impatient.
So it comes down to this – either:
1. I build a dual Opteron system based on 250s or 252s (and I’d probably wait for the 90nm “Troy” core) with 2 GB of DDR400ECC on a Tyan 2895 NF4 board, with 73 GB 15K UW320 SCSI, and 6800GTs in SLI.
2. I get an FX-55 with a Mach 2 GT, 1 GB DDR550 (TCCD), a good OC’able NF4 board, a X850XT PE and a single 74GB Raptor. My goal for this setup based on what I’ve seen of how the FX55s clock would be somewhere between 3.2 and 3.4 GHz prime stable. I don’t particularly care so much about benchmarking (be it SuperPI or anything else) – I want to be able to actually use this 24/7 for gaming and other things.
I have a number of questions, some of which I am sure have no good answers at this point, but I will ask anyway. Firstly, how realistic is the second option (basically, am I kidding myself)? Secondly, when I last OC’ed it was with water and with PATA/AGP. I’ve heard that SATA can cause some problems when it’s not locked (how do Raptors perform in this respect) and that similarly, NVidia GFX cards don’t do too well at high FSB as ATI because of the HSI bridges. I’d like to go with SLI, but I dunno if that would impact OCing to the speeds I’ve mentioned above and am leery of the HSI issue (though I am sure that no one has a good answer to this either since SLI is not available). Also, do I need to worry about a sub-ambient cooling solution for the north bridge at the above speeds? What would be a good motherboard to consider? I’ve heard good things about the DFI (huge voltage options), stability, etc – but I don’t know how it does at really high FSB with phase change (and it’s not out yet). NF4 is so new that it’s hard to tell…
I am also debating waiting for the FX-57 and the San Diego core, but then I have to buy sometime, and if I wait for FX-57 I might as well wait for the Toledo dual core FX chip. Any advice, including “shut up and buy!” would be appreciated greatly.
I’ve been without a decent computer and gaming platform ever since I sold my old 2.4B Northwood @ 3.3 to work, which was about 6 or so months ago. The intention then had been to wait for the FX-55 (which at that point I believed would be a San Diego cored 90nm CPU), get a Mach 2 GT and either do an SLI setup or whatever top-end card ATI had to offer at the time. It was a good plan… only reality had to complicate it! Damn. I find out that the FX-55 was to be based on a 130nm Sledgehammer part without SSE3, and that nVentiv went out of business; building my own phase change unit would be cheaper and a cool project, but I simply lack the time and tools. Furthermore, I didn’t really want a Vapochill LS, so I seriously began to reconsider my plans. I’d always wanted a dualie, not because I ever thought it would be better at single threaded games but because I could multi-task a lot better and could actually do productive things on it (sure a little gaming on the side never hurt anyone). Then, reality has to go and complicate things again… turns out the FX-55 clocks really well, and Mach 2 GTs are available again. So now, I’m faced with a terrible choice: abandon OCing (I’ve got no illusions about trying to OC Opterons on a Tyan board) in favor of SMP, or build essentially the original system I had planned, but with a very fast processor that will be less future proof. Let me explain the last comment: I think that with the advent of mainstream dual core CPUs application developers are going to have to start making multi-threaded apps and games. SSE3 and the other optimizations included in the rev E. core are also important to me. I know that the performance impact probably won’t be as significant as other factors, but if I am going to plop down as much money as this is going to cost, I want to be happy with it afterwards – and not feel like I missed out on something better by being impatient.
So it comes down to this – either:
1. I build a dual Opteron system based on 250s or 252s (and I’d probably wait for the 90nm “Troy” core) with 2 GB of DDR400ECC on a Tyan 2895 NF4 board, with 73 GB 15K UW320 SCSI, and 6800GTs in SLI.
2. I get an FX-55 with a Mach 2 GT, 1 GB DDR550 (TCCD), a good OC’able NF4 board, a X850XT PE and a single 74GB Raptor. My goal for this setup based on what I’ve seen of how the FX55s clock would be somewhere between 3.2 and 3.4 GHz prime stable. I don’t particularly care so much about benchmarking (be it SuperPI or anything else) – I want to be able to actually use this 24/7 for gaming and other things.
I have a number of questions, some of which I am sure have no good answers at this point, but I will ask anyway. Firstly, how realistic is the second option (basically, am I kidding myself)? Secondly, when I last OC’ed it was with water and with PATA/AGP. I’ve heard that SATA can cause some problems when it’s not locked (how do Raptors perform in this respect) and that similarly, NVidia GFX cards don’t do too well at high FSB as ATI because of the HSI bridges. I’d like to go with SLI, but I dunno if that would impact OCing to the speeds I’ve mentioned above and am leery of the HSI issue (though I am sure that no one has a good answer to this either since SLI is not available). Also, do I need to worry about a sub-ambient cooling solution for the north bridge at the above speeds? What would be a good motherboard to consider? I’ve heard good things about the DFI (huge voltage options), stability, etc – but I don’t know how it does at really high FSB with phase change (and it’s not out yet). NF4 is so new that it’s hard to tell…
I am also debating waiting for the FX-57 and the San Diego core, but then I have to buy sometime, and if I wait for FX-57 I might as well wait for the Toledo dual core FX chip. Any advice, including “shut up and buy!” would be appreciated greatly.