sisandra memory benchmarks, good or bad give your opinion

[wakedog102]

I just built a A64 3000+ system with 1 stick of 512MB PC433 Mushkin ram, running at 2:3:3:5. I know that current P4 boards run dual memory channels (at least 865 and 875 do) but their memory scores seem inordately high compaired to what I'm getting with my new setup in sisandra. Does sisandra just test theoretical memory performace as opposed to real world? I guess I'm wondering in case I've done something wrong with my memory setup and I fix it by tweaking the bios.

 

[Markfw]

What scores ? And mine looked low too, and people said ignore sandra and I did. How do you like the new machine for the games/programs you usually run ? That the real important question!!!

 

[JustStarting]

P4 is quad pumped, AMD is only dual. Dual channel is insignificant in comparison to this- maybe a couple hundred points in either rig.

 

[wakedog102]

Well, right now I'm in a transitionary phase. I have a A64 3000+ / MSI K8T800 and the Mushkin 512MB RAM. I've got a 60GB and a 40GB ATA100 hard drives (don't even know if they have cache) running from my old P3 1GHz computer and a Ti4200 128MB from my old comp as well. Compaired to that rig, the new is nice. Once things are loaded (games, photoshop, etc) the comp is very snappy. Loading times have been rather slow though, which prompted me to buy 2 SATA 120GB 8MB cache drives which I plan to RAID 0. Hopefully that will help the load times. Games such as Warcraft 3 are running beautifully smooth with the new processor, whereas in large battles on the P3, it was a little clunky. I've been having problems with my Ti4200 crashing (it is RMAed) and ordered a BFG 5900 which should be in next week. Once that and the hard drives are in I will be able to give a definative opinion on my new rig.

HOWEVER, compaired to my 2.6GHz Xeon (HT = on) at work, also has 1GB RAM, 120GB HD ATA(not sure if it's 100 or 133 or cache size), and a matrox G550 card it seems slow on load times, things like programs opening and start menu loading. I'm really hoping it's the hard drives, and not the RAM or some other factor, but I guess I will find out shortly. I don't regret my upgrade though. I'll gladly take a stock A64 3000+ which beats out a P4 3.0 / 3.2GHz chip than a P4 2.6 / 2.8 GHz overclocked all the time to get P4 3.0/3.2 like speeds.

-Nate

 

[computerpro3]

i find it funny that at sisoft sandra memory benchmark, when i boot at 3.2ghz 430mhz memory and run it I get 3200mb/s. After I use clockgen and set my memory to 500mhz 2-3-2-6 i get the same thing lol. I dont think it even runs a bench.

 

[DAPUNISHER]

It isn't called SiSuck for nothin' and it certainly doesn't account for the reduced latency the on-die memory controller provides from what I've heard. I haven't used SiSuck in ages so I have no idea first hand.

 

[Duvie]

Worthless is more like it.....


The only thing you need to know is how ppl went from single channel DDR at pc3200 with about 2900-3000 in sandra scores to dual channel and getting like 4700-4800 on same fsb chip(533's).....that is 60% increase in performance...However real world test show maybe a 10-15% increase at best in certain tests......

Just remember not real world.....but also not theoretical since 2 stickd of pc3200 should yield theoretical numbers of 6.4gb/s or 6400 scor eyet that does not happen either....


I think most agree that memtest bandwidth and aida are a bit better and more accurate.