D
Dale
Nope. I started with one gig - a pair of 512MB sticks and doubled by adding
a second pair of 512MB sticks.
Dale
a second pair of 512MB sticks.
Dale
B
BSchnur
I got a 4.5 with one gig. I may well upgrade
and that the card has good heat handling as well. Also note that as
you move up the ladder with the GPU and memory on the card, you might
also need to take care regarding your power supply.
Look into x1300 or x1600 -- make sure your case has adequate cooling
and that the card has good heat handling as well. Also note that as
you move up the ladder with the GPU and memory on the card, you might
also need to take care regarding your power supply.
B
BSchnur
If you say it increased your score to 5.6, I believe you, but I'm confused.
It might well if the original memory was a single module and the
upgrade was to two match modules using the DDR2 doubled data rate.
It might well if the original memory was a single module and the
upgrade was to two match modules using the DDR2 doubled data rate.
K
Ken Blake, MVP
BSchnur said:
Thanks. Perhaps that's it.
D
Dale
I've heard the same results as mine from other posters: More RAM - higher
score.
As my other post said, I started out with 2 sticks of 512MB in a pair and
added a second pair of 2 sticks of 512MB.
score.
As my other post said, I started out with 2 sticks of 512MB in a pair and
added a second pair of 2 sticks of 512MB.
K
Ken Blake, MVP
Dale said:
Yes, I wasn't denying the result you reported, just trying to understand it.
On the face of it, it doesn't seem like it should be.
In that case, what BSchnur suggests below would not seem to be correct.
B
BSchnur
I've heard the same results as mine from other posters: More RAM - higher
bit. I've seen some boards that have four slots and don't make it
clear how the pairs are oriented, so that if one pair isn't placed
properly (and re-checked at POST), when you added a second pair, you
actually 'paired' the original two single modules.
That being said, it is apparent that Vista will be 'happier' (ie
perform better) with 2G versus 1G.
Over in the Novell newsgroups, we make the some sort of suggestion,
NetWare 6.5 (like Vista) really likes seeing 2G of memory compared to
1G of memory for performance. Like Vista NetWare 6.5 has a minimum
memory requirement of 512M and like Vista, running down there is a
recipe for performance bottlenecks.
OK -- I'm assuming that the first pair was installed to operate at 128
bit. I've seen some boards that have four slots and don't make it
clear how the pairs are oriented, so that if one pair isn't placed
properly (and re-checked at POST), when you added a second pair, you
actually 'paired' the original two single modules.
That being said, it is apparent that Vista will be 'happier' (ie
perform better) with 2G versus 1G.
Over in the Novell newsgroups, we make the some sort of suggestion,
NetWare 6.5 (like Vista) really likes seeing 2G of memory compared to
1G of memory for performance. Like Vista NetWare 6.5 has a minimum
memory requirement of 512M and like Vista, running down there is a
recipe for performance bottlenecks.
K
Ken Blake, MVP
BSchnur said:
Sure. That's never been at issue. My only question had to do with the
Windows Vista "score," which Vista defines as "Memory operations per
second." I asked "Why should doubling the amount of RAM have any effect on
the memory speed?"
Ken
D
Dale
Nope. You assume wrong. I understand the pairing of memory on the
motherboard and, along with that knowledge, followed the instructions in the
book. How about it is just something you don't understand? For whatever
reason, the results are what they are. You're just wrong. Get over it.
Dale
motherboard and, along with that knowledge, followed the instructions in the
book. How about it is just something you don't understand? For whatever
reason, the results are what they are. You're just wrong. Get over it.
Dale
F
Frankster
Ken Blake said:
I do see your point.
It could be that since you have twice as much memory, you can accommodate
twice as many "operations" at one time, leading to twice as many operations
being completed in the same amount of time. Doubling your "speed" of
operations completed. Or, another way of putting it, completing a specified
number of operations in half the time.
Well... anyway, I just made that up. But I was unable to find anything more
comprehensive. And it does sorta make sense.
-Frank
B
BSchnur
Understood -- I guess for me, I'd look at the Sandra scores instead of
Microsoft to get more precision on this.
I will also check out the score differences when I move from a matched
pair of 512m modules to a matched pair of 1G modules in one of my Vista
systems in the near future.
B
BSchnur
Dale, OK here is what I posted:
Here is your reply:
So, aside from your natural inclination to start yet another food
fight, are you saying what you posted -- that you understand how to get
the 128 bit matched pair memory handling, but in order to tell me that
I assumed wrong, you deliberately installed them according to the book
to get 64 bit? That's going a bit far.
Here is your reply:
So, aside from your natural inclination to start yet another food
fight, are you saying what you posted -- that you understand how to get
the 128 bit matched pair memory handling, but in order to tell me that
I assumed wrong, you deliberately installed them according to the book
to get 64 bit? That's going a bit far.
D
Dale
No. I am not starting a food fight. I have posted over and over again and
you keep looking for assumptions you can make to support a theory that you
have created in your own mind about what the results should be.
I am telling you the facts of what the results are, not what they should be.
You can argue them all you want but you're only making it up. You haven't
done the test. You really don't know much of which you speak.
Dale
you keep looking for assumptions you can make to support a theory that you
have created in your own mind about what the results should be.
I am telling you the facts of what the results are, not what they should be.
You can argue them all you want but you're only making it up. You haven't
done the test. You really don't know much of which you speak.
Dale
B
BSchnur
OK -- I'll admit I'm confused. I have difficulty tracking the three
posts regarding my assumption on dual channel memory and performance.
That said, perhaps you have test reports for the same configurations
(2G and then 4G) in both Vista 32 and Vista 64. I'd be interested in
your reports using the Microsoft tests.
posts regarding my assumption on dual channel memory and performance.
That said, perhaps you have test reports for the same configurations
(2G and then 4G) in both Vista 32 and Vista 64. I'd be interested in
your reports using the Microsoft tests.
K
Ken Blake, MVP
BSchnur said:
Thanks. I'd be interested in hearing about your results.
This is obviously not of earthshaking significance, but I'm curious. I
suspect that Vista's definition of "Memory operations per second" may just
be incorrect.
B
BSchnur
Ken, I am expecting to get a memory matched pair of Kingston DDR2 667
1G modules this week. What I hope to do is use them to replace a pair
of memory matched DDR2 533 512M modules on a Vista system. So I will
have two variables -- memory module speed and amount of memory. I will
run both the Sandra memory benchmarks and the Microsoft test before and
after and will post results. Might be this coming weekend.
1G modules this week. What I hope to do is use them to replace a pair
of memory matched DDR2 533 512M modules on a Vista system. So I will
have two variables -- memory module speed and amount of memory. I will
run both the Sandra memory benchmarks and the Microsoft test before and
after and will post results. Might be this coming weekend.
K
kjk
We await your results. Thanks.
K
Ken Blake, MVP
BSchnur said:
Thanks.
B
BSchnur
So I was in a hurry and did this test on a somewhat older test bed --
AMD 939X2 3800 -- with DDRAM though one which supports matched pairs.
I added a second pair of PC3200 512M modules -- thus going from 1G to
2G of RAM -- installation is with RC1 (therefore Vista Ultimate) 64
bit. The RAM performance changed -- from 4.5 to 4.6.
Not in the earthshaking department.
As I noted, I'm waiting on a pair of DDR2 1G 667 modules -- those will
go on a system which only has memory slots and will replace a pair of
DDR2 512M modules which are 533's -- and also show a RAM performance
number of 4.5 in Microsoft speak. I would expect at least a marginally
bigger change as the actual memory speed bumps up from 533 to 667.
(from PC2 4200 to PC2 5300).
AMD 939X2 3800 -- with DDRAM though one which supports matched pairs.
I added a second pair of PC3200 512M modules -- thus going from 1G to
2G of RAM -- installation is with RC1 (therefore Vista Ultimate) 64
bit. The RAM performance changed -- from 4.5 to 4.6.
Not in the earthshaking department.
As I noted, I'm waiting on a pair of DDR2 1G 667 modules -- those will
go on a system which only has memory slots and will replace a pair of
DDR2 512M modules which are 533's -- and also show a RAM performance
number of 4.5 in Microsoft speak. I would expect at least a marginally
bigger change as the actual memory speed bumps up from 533 to 667.
(from PC2 4200 to PC2 5300).
D
Dale
It's a nearly pointless comparison. My PC is a Core 2 DUO 6400 with,
originally, 1 GB of DDR2 800 MHZ, not DDR memory.
That said, your comparison did show a change. Your posts before said there
would be no change. Why would your old testbed PC change even a little if
your totally uneducated guess about how the test works was correct?
I don't particularly care what your next test shows either. Basically
you're just calling me a liar, I guess. You seem to be on a mission to
prove that you know what you're talking about and that I am wrong.
What gall you have to think the world waits for your "real test" to show the
truth of the matter as if you have some assumed level of credibility here
and until BSchnur says it, we just can't know what's real and what's not.
I told you what my results were. If you don't like them, do your own test.
It doesn't change my results.
Dale
originally, 1 GB of DDR2 800 MHZ, not DDR memory.
That said, your comparison did show a change. Your posts before said there
would be no change. Why would your old testbed PC change even a little if
your totally uneducated guess about how the test works was correct?
I don't particularly care what your next test shows either. Basically
you're just calling me a liar, I guess. You seem to be on a mission to
prove that you know what you're talking about and that I am wrong.
What gall you have to think the world waits for your "real test" to show the
truth of the matter as if you have some assumed level of credibility here
and until BSchnur says it, we just can't know what's real and what's not.
I told you what my results were. If you don't like them, do your own test.
It doesn't change my results.
Dale