Go from 1 GB to 2 GB RAM, feel speed increase?

[WhipperSnapper]

Not exactly a CPU or overclocking issue, but it relates to speed, our favorite topic.

I jumped on that $50 + shipping after rebate deal for 2 x 512 Corsair Value RAM at Outpost and popped it in, giving me 4 x 512 of the Corsair Value RAM.

I didn't think it would really make a difference, but I think my system feels faster and snappier now and all I've been doing is Internet surfing and MS Office applications. Does it make any sense that I would really notice a speed difference? Could my system have enjoyed a real speed increase, or is it just my imagination?

Processor is an Opteron 148 overclocked to 2.75 Ghz on an Epox EP-9NPA+Ultra motherboard running WinXP Pro SP2.

 

[Vegitto]

In your case: no. By using 4 memory slots instead of 2, you're running at 2T, which immediately snoops away the (possible) advantage you had by having 2GB instead of just 1(GB ).

 

[WhipperSnapper]

Yeah, I was wondering about that. I don't think being at 2T has hurt me at all though. At least it doesn't feel that way. The biggest bottleneck in my system no is my hard drives.

Is there a way to see whether you're at 1T or 2T in CPUZ?

 

[Ricemarine]

Originally posted by: WhipperSnapper

Yeah, I was wondering about that. I don't think being at 2T has hurt me at all though. At least it doesn't feel that way. The biggest bottleneck in my system no is my hard drives.

Is there a way to see whether you're at 1T or 2T in CPUZ?

Nope, but you can see it in A64 Tweaker 0.6 beta.

 

[edmundoab]

not willing to invest in 2 sticks of 2GB RAM?

 

[TrevorRC]

Originally posted by: Vegitto
In your case: no. By using 4 memory slots instead of 2, you're running at 2T, which immediately snoops away the (possible) advantage you had by having 2GB instead of just 1(GB ).

Vegitto, it's been proven that 2T only causes a 3-5% performance decrease.

Check AngryGames' thread over at DFI.

Originally posted by: WhipperSnapper

Yeah, I was wondering about that. I don't think being at 2T has hurt me at all though. At least it doesn't feel that way. The biggest bottleneck in my system no is my hard drives.

Is there a way to see whether you're at 1T or 2T in CPUZ?

Same here..
I can feel windows booting up, the entire system SCREAMING at my hard drives.

 

[Bobthelost]

Still a 3-5% hit is a 3-5% hit. For most of your time you won't break 1GB so any speed increases are purely in the mind. As in reality your computer will be performing a few % worse.

However you will see a fair bit of improvement for gaming

 

[nealh]

I thought the same thing when moving from 1GB to 2GB and I do thing things feel faster ..can not explain it

 

[TrevorRC]

Originally posted by: Bobthelost
Still a 3-5% hit is a 3-5% hit. For most of your time you won't break 1GB so any speed increases are purely in the mind. As in reality your computer will be performing a few % worse.


However you will see a fair bit of improvement for gaming

Bob, it's a 3-5% hit AFTER a 100% increase.



He doubles his RAM, thereby unbottlencking any thing that was starved, then takes a 3-5% hit. Unless anything comes up to 100% RAM utilization with 2GB, he'll never know the 3-5% difference (the 3-5% difference ONLY applies when comparing 2x1GB to 4x512MB. It's not a 3-5% decrease in total performance. Think about it. Why would, logically speaking, doubling your RAM DECREASE the performance?0

 

[DeathReborn]

Even going from Geil Value 1GB (2x512MB) BH-5 to the 2x1GB OCZ I have now isn't noticeable in most situations. Okay so BF2 loads quicker, not a big deal since I hardly play it. In my case there was not much reason to upgrade, I just did cos it was cheap.

 

[furballi]

Guarantee speed drop when you run 4 x 512MB because the CPU memory controller must drop down from 1T timing to 2T. You can verify this by running 1M digits Super Pi with two, then four sticks of RAM. Your calculation time will probably drop by 1.5 second. That's about a 4% speed penalty ALL THE TIME. Even if you can set your system to be stable at 1T with four modules, you will still not see any improvement UNLESS you're running multiple applications requiring more than 1GB of RAM.

From a design stand point, having too much unused RAM is a waste of $. It costs a few extra CPU clock cycles to manage the extra RAM, plus you're adding more heat load to the case.

Repost your 1M digits SP times with 2 and 4 modules here.

 

[Bull Dog]

Originally posted by: furballi
Guarantee speed drop when you run 4 x 512MB because the CPU memory controller must drop down from 1T timing to 2T. You can verify this by running 1M digits Super Pi with two, then four sticks of RAM. Your calculation time will probably drop by 1.5 second. That's about a 4% speed penalty ALL THE TIME. Even if you can set your system to be stable at 1T with four modules, you will still not see any improvement UNLESS you're running multiple applications requiring more than 1GB of RAM.

From a design stand point, having too much unused RAM is a waste of $. It costs a few extra CPU clock cycles to manage the extra RAM, plus you're adding more heat load to the case.

Repost your 1M digits SP times with 2 and 4 modules here.

You are aware that there are other programs in the world besides SuperPI right?

 

[Madwand1]

Originally posted by: WhipperSnapper
I think my system feels faster and snappier now and all I've been doing is Internet surfing and MS Office applications. Does it make any sense that I would really notice a speed difference? Could my system have enjoyed a real speed increase, or is it just my imagination?

More RAM will give you a large file system cache, which is likely to be useful over the long run and while using multiple applications. So your perceived benefit could be from that.

If you're too lazy to shut down, power down, uninstall the RAM, and reverse (as I usually am), you could use the boot.ini switch /maxmem=1024 to simulate going down to 1 GB of RAM. This will not change hardware effects, including 1T/2T, but will show you the software effects, if any, due to additional / reduced RAM.

If you want to check the effects of the system cache directly (and this is advisable if you're doing any disk testing for example), you can use the Sysinternals freeware CacheSet.

 

[Hacp]

Originally posted by: furballi
Guarantee speed drop when you run 4 x 512MB because the CPU memory controller must drop down from 1T timing to 2T. You can verify this by running 1M digits Super Pi with two, then four sticks of RAM. Your calculation time will probably drop by 1.5 second. That's about a 4% speed penalty ALL THE TIME. Even if you can set your system to be stable at 1T with four modules, you will still not see any improvement UNLESS you're running multiple applications requiring more than 1GB of RAM.

From a design stand point, having too much unused RAM is a waste of $. It costs a few extra CPU clock cycles to manage the extra RAM, plus you're adding more heat load to the case.

Repost your 1M digits SP times with 2 and 4 modules here.

Wrong! You can def run 4x 512 and still get 1T. You just either need the right motherboard, or the right memory,

 

[bjc112]

I noticed ZERO performance gains, when less than 2GB was needed.

 

[markkleb]

I have 4 one gig sticks of PQI turbo 400 lat2 with win64 and it is much faster than 1 gig. I also upgraded to 2 segate 80 gig satas in raid 0, also a big improvement.

I noticed a big jump from my Vic 20. LOL

I used to buy 512's but now 1 gigs are as cheap now. What a cool world we live in.

 

[TrevorRC]

Originally posted by: Bull Dog

Originally posted by: furballi
Guarantee speed drop when you run 4 x 512MB because the CPU memory controller must drop down from 1T timing to 2T. You can verify this by running 1M digits Super Pi with two, then four sticks of RAM. Your calculation time will probably drop by 1.5 second. That's about a 4% speed penalty ALL THE TIME. Even if you can set your system to be stable at 1T with four modules, you will still not see any improvement UNLESS you're running multiple applications requiring more than 1GB of RAM.

From a design stand point, having too much unused RAM is a waste of $. It costs a few extra CPU clock cycles to manage the extra RAM, plus you're adding more heat load to the case.

Repost your 1M digits SP times with 2 and 4 modules here.

You are aware that there are other programs in the world besides SuperPI right?

I don't think he does.

Anyway, the point is:
Windows will load up and shut down faster--less information is stuck on a slower hard drive.
Applications are less likely to be cached on the hard drive (when there is more availible RAM), and as such, things will run substantially smoother.

 

[Stoneburner]

When i went from 1 gig at ddr400 and 2t to 2 gig ddr400 at 1t i noticed much better windows general application responsiveness, everything from windows explorer to firefox. Then I went to ddr500 at 2t and it feels slower.

Moral of the story.... 1t seems to be useful.

 

[fireontheway]

Got this last nite,very happy .

 

[koolzb]

I read many review suggested to use 2 stick RAM either 2x256 or 2x512 or 2x1024 to get maximum speed of the system.
Bencmark shows clearly that 4 slot RAM use will drop performance significantly.
Cannot remember where it was reviewed.

 

[OVerLoRDI]

In my experience going from 1gb of ram to 2gb was actually a nice increase. Games loaded faster, closed faster, and I was able to use more detailed textures without having to page the hard drive. Its not quite night and day like going from 512mb to a gig but I dont regret spending the money. Also I'm using 2 x 1gb sticks so I still have T1.

 

[WhipperSnapper]

I have the Epox EP-9NPA+Ultra and I'm wondering if I could set it to 1T. Earlier today I was in the BIOS and it listed the RAM setting as Auto/2T, which makes me wonder what would happen if I set it to 1T.

What would happen? Would my system just freeze up? I'll have to give it a go on my next reboot.

 

[furballi]

It may not load at 1T (reset CMOS if lock up). The system may not pass Prime95 and/or Memtest at 1T. Again, I'm sure that you will not be able to run 1T with your rig. If you must default to 2T, then use Super Pi 1M digits as a quick test to check for speed.

 

[xtknight]

Originally posted by: Vegitto
In your case: no. By using 4 memory slots instead of 2, you're running at 2T, which immediately snoops away the (possible) advantage you had by having 2GB instead of just 1(GB ).

Actually, it hardly makes a difference. 2 GB/4 sticks is still worlds faster with everything you can imagine (except those theoretical memory bandwidth tests).

I can kill 39 IE windows in a group in one second. Now I call that a serious lack of swapping.

 

[koolzb]

I second xtknight,
2x 512MB reported to give best performance on most applications comparing to 2x 1024 sticks.