August 13, 200916 yr Hi everybody, I have this week just upgraded my computer system for running FSX, after carefull consideration I have purchased the following components1. i7 975 (o/c 4.2 stable test ottc 80c 1hr ok)2. 6gig ddr3 2000mhz corsair doms 8-8-8-243. western digital 10,000rpm h/drive4. 1500watts power supply unit5. BfG 295 watercooled g/card6. zalman reserator xt watercooling system7. silverstone tj case8. Asus P6t Deluxe m/boardAll built up now and ready to start loading fsx on but before I do guys i am looking for some advice Please,I have managed to overclock to 4.2 quite easy by just increasing my vcore to1.35volts.ran occt for 1 hour , temps got to 79/81c but passed test.4.4ghz cpu too hot!zalman reserator running on cpu & v/card for cooling.not loaded fsx on yet because reading on your site people are saying memory timings for best performance should be 6-6-6 or 7-7-7 will my exspensive 2000mhz dominator ddr3 8-8-8-24 do the job or will I have to shoot for 7-7-7 timings with them. any advice on this system please advice me as Iam new to overclocking and all your advice would be greatfull (specialy NICKN ). Also any body read about the NEW case from Thermaltake EXPREESAIR it as a mini refrigerated built in to the case and claims to run 20c lower temps than liquid cooling,any body any thought on that ! Thanks Andyall
August 13, 200916 yr The point of lower timings is to reduce memory access latency. Timings in and of themselves do not define latency, it is a function of clockspeed as well. Higher clockspeed makes up for higher timings. Your DDR3 2000 8-8-8 RAM would be equivalent to DDR2 1000 4-4-4 in access latency. Your RAM's minimum access time is 8ns, quite fast already.
August 13, 200916 yr Author The point of lower timings is to reduce memory access latency. Timings in and of themselves do not define latency, it is a function of clockspeed as well. Higher clockspeed makes up for higher timings. Your DDR3 2000 8-8-8 RAM would be equivalent to DDR2 1000 4-4-4 in access latency. Your RAM's minimum access time is 8ns, quite fast already.thankyou for that reply mate, what would you do stick with the 2000mhz ddr3 domms or go for muskin redline 1600mhz 6-6-5-18 ?
August 13, 200916 yr Moderator thankyou for that reply mate, what would you do stick with the 2000mhz ddr3 domms or go for muskin redline 1600mhz 6-6-5-18 ?Andy - don't have the exact link but NickN was discussing this a while back and if IIRC, he flat stated that the Mushkin 1600 6-6 is the way to go. You might want to do a bit of searching just to be sure - I think it was on Simviation.Vic RIG#1 - I9 14900K MSI Pro z790 RTX 5070Ti 40" 4K Monitor 3840x2160
August 14, 200916 yr If you can clock your RAM to the full 2GHz I'd say go that route. The latency difference between DDR3 1600 6-6-5 and DDR3 2000 8-8-8 is tenths of a nanosecond, but the DDR3 2000 will have an ~25% bandwidth advantage.
August 14, 200916 yr If you can clock your RAM to the full 2GHz I'd say go that route. The latency difference between DDR3 1600 6-6-5 and DDR3 2000 8-8-8 is tenths of a nanosecond, but the DDR3 2000 will have an ~25% bandwidth advantage.Andy,http://forums1.avsim.net/index.php?showtopic=241243About two-thirds of the way down the threadDDR3 2000 8-8-8 is SLOWER than DDR3 1600 6-6-5 Especially in FSX.Don't buy into the DDR3 2000 unless it's 7-7-7. And even then it's easier and more reliable to run 1600 6-6-5, voltages are lower on the QPI. Be very careful with any of these Elpida Hyper chip DIMMs. They're dropping like flies these days even at 1.65v. They are the fastest things that's around but not very reliable yet. If you do get into trouble Mushkin WILL honor the warranty though. Those Redlines are THE ticket right now.-jk
August 14, 200916 yr Andy,http://forums1.avsim.net/index.php?showtopic=241243About two-thirds of the way down the threadDDR3 2000 8-8-8 is SLOWER than DDR3 1600 6-6-5 Especially in FSX. Nothing in that thread backs up this claim. Trading off 25% bandwidth gain for a percent or two latency difference doesn't make sense, especially in light of this thread which demonstrates this very fact. Be very careful with any of these Elpida Hyper chip DIMMs. They're dropping like flies these days even at 1.65v. They are the fastest things that's around but not very reliable yet. If you do get into trouble Mushkin WILL honor the warranty though. Those Redlines are THE ticket right now.Corsair recently re-launched the Dominator GT line after getting together with Elpida to fix the problem. In fact, they just expanded the line to include new dual-channel kits approved for AMD Phenom II systems.
August 14, 200916 yr Andy,Go to that thread that Max just posted and read down to the timings chart than Nick put in: CAS latency (ns) vs. freq (MHz). The take home message is that timings DO matter and if you get 1600 C6 you will be better off than 2000 C8. That small latency difference will do more in FSX than any bandwidth hype. There's plenty of bandwidth either way. And remember FSX cannot be measured by FPS benchmarks alone. -jk Corsair recently re-launched the Dominator GT line after getting together with Elpida to fix the problem. In fact, they just expanded the line to include new dual-channel kits approved for AMD Phenom II systems.That doesn't mean the problem's fixed yet. Only time will tell.-jk
August 14, 200916 yr Andy,Go to that thread that Max just posted and read down to the timings chart than Nick put in: CAS latency (ns) vs. freq (MHz). The take home message is that timings DO matter and if you get 1600 C6 you will be better off than 2000 C8. That small latency difference will do more in FSX than any bandwidth hype. There's plenty of bandwidth either way. And remember FSX cannot be measured by FPS benchmarks alone.The benchmarks the creator of that thread has provided demonstrate a clear trend that performance scales with bandwidth. Read my first post in that thread and it is explained clearly. Nick's latency graph looks fancy but doesn't mean anything WRT FSX performance. It really comes down to this: raw performance numbers from the actual application versus hypothetical postulations. I know which one I'd trust. That doesn't mean the problem's fixed yet. Only time will tell.-jkI understand your caution but I also trust Corsair's engineers have sorted things out. Perhaps all the RAM in the channel is not yet of the "fixed" variety but I do believe they have indeed fixed the issues with Elpida hyper ICs.
August 14, 200916 yr Hi,I hope they are fixed, took them four days to give me an RMA for my 7-8-7-20 sticks. Time to overnight these to Corsair. I understand your caution but I also trust Corsair's engineers have sorted things out. Perhaps all the RAM in the channel is not yet of the "fixed" variety but I do believe they have indeed fixed the issues with Elpida hyper ICs. Former Beta Tester - (for a few companies) - As well as provide Regional Voice Set Recordings Two: AMD-9950X | One: AMD-7950X3D | Three: Asus TUF 4090s | Three: 64GB DDR5 RAM 6000mhz | Three: Cosair 1300 P/S | Three: 990Pro 2TB NVME One: Eugenius ECS2512 - 2.5 GHz Switch | Three: Ice Giant Elite CPU Coolers | Three: 75" 4K UHDTVs | One: Boeing 737NG Flight Deck
August 14, 200916 yr Hi,I hope they are fixed, took them four days to give me an RMA for my 7-8-7-20 sticks. Time to overnight these to Corsair. http://www.madshrimps.be/?action=getarticl...amp;articID=928Looks like they're not using hypers anymore. Still Elpida, just not hypers. And the (limited) testing looks good.I hope that's the last time you have to do that, Mike.If I could run 2000 CAS 7, on my poor C0, I would.-jk
Create an account or sign in to comment