Get a gigabyte of Crucial's RAM (two 512MB sticks with Apple-style heatsinks) for $200. There's also a few other places offering it now such as TransIntl:
Crucial's pricing is pretty good, but they don't have 1GB sticks yet.
That other link is good, but their prices are high.
I know I'm picky, but newegg is selling 1GB FB-DIMMs for $169. They don't have the Mac Pro heatsinks--but come on, a heatsink is a $10 part, if that. I'm looking for about $350-360 for a 2 x 1GB kit.
Wow, the people from Anandtech are fucking liars, they claimed it would only increase performance by 3%.
Most of the stuff I've read at Anandtech seems balanced, but for whatever reason, I trust the BareFeats site more for mac-specific tests. \
When Rob-Art makes a mistake, he owns up to it quickly on the site. And in this case, I agree with him - "it can't hurt" both in terms of the mem bus and spreading out the heat sources.
Yeah, and I need 2GB regardless of channel saturation, so it's win-win.
Wait, was he testing two gig sticks (two channels) versus four 512 sticks (four channels), or something less fair where the four-channel setup had more RAM? Because that could have skewed it a bit.
Wait, was he testing two gig sticks (two channels) versus four 512 sticks (four channels), or something less fair where the four-channel setup had more RAM?
He had 6 GB, but didn't use them in every test.
Quote:
Originally Posted by Bare Feats
We had four 512MB modules and four 1GB modules. The three combinations we tried using both risers produced a fill rate 34% faster than the five combinatons we tried using one riser card...
If you ask me, they should be testing what I'd call typical - ie. configuations less than 4 GB, more preferably, 2 GB (ie. is 4x512 better than 2x1024?). I suppose this brings other variables into the picture (HD performance, as an example) but given that we're using the same systems, I think the results would still be valid.
Yeah, it is pretty damned confusing because then we have to thing of whether having larger stick sizes results in longer memory reference times, etc etc.
I'll just stick with the ol' rule of thumb: MORE RAM
4x512 is better than 2x1024. An interesting question is whether 8x512 is worse than 4x1024 (I think it is because of latency issues on the "back" RAM).
4x512 is better than 2x1024. An interesting question is whether 8x512 is worse than 4x1024 (I think it is because of latency issues on the "back" RAM).
My guess, and it's purely a guess, is that the gains from saturating the bus is an order of magnitude more than the potential gains from avoiding this latency.
My guess, and it's purely a guess, is that the gains from saturating the bus is an order of magnitude more than the potential gains from avoiding this latency.
4 properly placed DIMMs saturate the bus (it's quad-channel).
4 properly placed DIMMs saturate the bus (it's quad-channel).
I edited my post to make it more clear. I don't think the latency is anything a user could notice, versus the percentage level gains acheived by saturating the bus. Just found another good link:
I just got an email from Apple informing me that the price of my Mac Pro had dropped by $90. Shipping date is still holding at September 19th. Below are the specs. I don't know which component saw the price reduction.
I just got an email from Apple informing me that the price of my Mac Pro had dropped by $90. Shipping date is still holding at September 19th. Below are the specs. I don't know which component saw the price reduction.
2.00 GHz Quad Xeon
1GB 667 DDR2 FB DIMM ECC-2x512
ATI Radeon X1900 XT 512MB
250GB Serial ATA 3Gb/s drive
Airprt Extrm & BT 2.0+EDR
I have the same config, and I got a price drop too. (Mine was $100. Yours was $90?) That's nice.
Comments
If you have the 3.0 gHz, how about giving us a benchmark on the threadedFactorialHighRes app?
See if it can beat 89 seconds for 2 billion. That is what the 2.66 Woodcrest gets.
78.71 seconds.
If I had more memory (to fill out all the channels), it would probably be a little faster.
Get a gigabyte of Crucial's RAM (two 512MB sticks with Apple-style heatsinks) for $200. There's also a few other places offering it now such as TransIntl:
http://www.transintl.com/store/categ...estTimeOut=500
and OWC.
Crucial's pricing is pretty good, but they don't have 1GB sticks yet.
That other link is good, but their prices are high.
I know I'm picky, but newegg is selling 1GB FB-DIMMs for $169. They don't have the Mac Pro heatsinks--but come on, a heatsink is a $10 part, if that. I'm looking for about $350-360 for a 2 x 1GB kit.
Crucial says they'll have gigsticks within 1-2 weeks.
For me, 2 GB will be plenty, and 4x512 is the best way to saturate the memory bus. 8)
Of course, that means I'll have to play more UT2004, 'cause Barefeats sez that's the only time I'll notice the difference.
http://www.barefeats.com/quad09.html
For me, 2 GB will be plenty, and 4x512 is the best way to saturate the memory bus. 8)
Of course, that means I'll have to play more UT2004, 'cause Barefeats sez that's the only time I'll notice the difference.
http://www.barefeats.com/quad09.html
so drunk but still thinking about my fukinvg macpro
For me, 2 GB will be plenty, and 4x512 is the best way to saturate the memory bus. 8)
Of course, that means I'll have to play more UT2004, 'cause Barefeats sez that's the only time I'll notice the difference.
http://www.barefeats.com/quad09.html
Wow, the people from Anandtech are fucking liars, they claimed it would only increase performance by 3%.
Wow, the people from Anandtech are fucking liars, they claimed it would only increase performance by 3%.
Most of the stuff I've read at Anandtech seems balanced, but for whatever reason, I trust the BareFeats site more for mac-specific tests.
When Rob-Art makes a mistake, he owns up to it quickly on the site. And in this case, I agree with him - "it can't hurt" both in terms of the mem bus and spreading out the heat sources.
Wait, was he testing two gig sticks (two channels) versus four 512 sticks (four channels), or something less fair where the four-channel setup had more RAM? Because that could have skewed it a bit.
Wait, was he testing two gig sticks (two channels) versus four 512 sticks (four channels), or something less fair where the four-channel setup had more RAM?
He had 6 GB, but didn't use them in every test.
We had four 512MB modules and four 1GB modules. The three combinations we tried using both risers produced a fill rate 34% faster than the five combinatons we tried using one riser card...
If you ask me, they should be testing what I'd call typical - ie. configuations less than 4 GB, more preferably, 2 GB (ie. is 4x512 better than 2x1024?). I suppose this brings other variables into the picture (HD performance, as an example) but given that we're using the same systems, I think the results would still be valid.
I'll just stick with the ol' rule of thumb: MORE RAM
4x512 is better than 2x1024. An interesting question is whether 8x512 is worse than 4x1024 (I think it is because of latency issues on the "back" RAM).
My guess, and it's purely a guess, is that the gains from saturating the bus is an order of magnitude more than the potential gains from avoiding this latency.
The representative I called said that the earliest they're anticipating X1900 XT shipments is the 30th.
So two days... and its the big day... omg please!
My guess, and it's purely a guess, is that the gains from saturating the bus is an order of magnitude more than the potential gains from avoiding this latency.
4 properly placed DIMMs saturate the bus (it's quad-channel).
4 properly placed DIMMs saturate the bus (it's quad-channel).
I edited my post to make it more clear. I don't think the latency is anything a user could notice, versus the percentage level gains acheived by saturating the bus. Just found another good link:
http://www.macintouch.com/reviews/ma...up.html#memory
2.00 GHz Quad Xeon
1GB 667 DDR2 FB DIMM ECC-2x512
ATI Radeon X1900 XT 512MB
250GB Serial ATA 3Gb/s drive
Airprt Extrm & BT 2.0+EDR
I just got an email from Apple informing me that the price of my Mac Pro had dropped by $90. Shipping date is still holding at September 19th. Below are the specs. I don't know which component saw the price reduction.
2.00 GHz Quad Xeon
1GB 667 DDR2 FB DIMM ECC-2x512
ATI Radeon X1900 XT 512MB
250GB Serial ATA 3Gb/s drive
Airprt Extrm & BT 2.0+EDR
I have the same config, and I got a price drop too. (Mine was $100. Yours was $90?) That's nice.
I have the same config, and I got a price drop too. (Mine was $100. Yours was $90?) That's nice.
I bought from the Ed store.