Q1. Lets put it this way. HP themselves have bundles with 2 slots occupied as their "recommended" configurations.
A1. Yes, that's correct. However this doesn't fully utilize the RAM bandwidth available to a Xeon 5500 series processor as compared to Triple Channel DDR3. As with the move to Dual Channel RAM back in 2002, Triple Channel will result in a performance improvement over Dual Channel. Admittedly, it isn't as much of a performance improvement (due mainly to the newer cache architectures in modern CPUs), but it is still an improvement. So, if you want the fastest performance from a particular type of RAM, Triple Channel will be faster than Dual Channel will be faster than Single Channel.
Q2. I think the triple channel only works with twin CPU setups (ie. Not with one CPU) AFAIK
A2. No, that's not correct. In multiple-socket motherboards, half of the RAM slots (in a Triple Channel bank) are available with one CPU installed and the other half with the second CPU installed. Obviously for 4-way systems, 25% of the RAM slots will be available per CPU installed.
Q3. Won't DDR-1600 always be faster than DDR-1033?
A3. No. Not only does the memory speed matter, but the latency matters. The latency often matters more than the overall memory speed as it results in faster retrievals of small quantities of data, whereas a faster overall clockspeed will normally result in faster transfer of larger quantities of data. If you want to have a good look at the differences between RAM clock speeds and latencies, have a read of this Tom's Hardware article.
Q4. Aren't the Core i7 CPUs really just a Core 2 Quad that's been stuck into one package instead of a pair of Core 2 Duos glued together?
A4. No. The Core i7/Xeon 5500 series CPU core and cache architecture have been redesigned. There's significant differences between the way a Core 2 and a Core i7 CPU work, not just the doubling of the cores on the one die.
Q5. Is a Core i7 with DDR3 really faster than a Core 2 Quad with DDR2?
A5. Yes. A comparison of a Core i7 920 (the slowest one available - 2.66 GHz, 8 MB L2 Cache) with a Core 2 Quad Q9400 (2.66 GHz, 6 MB L2 Cache) or a Core 2 Quad Q9550 (2.83 GHz, 12 MB L2 Cache) will likely result in a 10% or more increase in speed. Of course that will depend on the applications you use, however if they are modern apps (multi-threaded) then you'll see around this or more. So, at the same clock speed, you'll see a performance improvement.
Q6. Is it true that Hyper-Threading makes a triumphant return in the Core i7/Xeon 5500?
A6. Yes and no. Xeon processors have always had Hyper-Threading (since it was introduced) however the Core i7 definitely heralds the Return of Hyper-Threading in the desktop processor space. :)
If there's anything that's still not clear (and I don't mean propellor head clear, just a quick answer to a quick question) leave a comment and I'll reply.
Regards,
The Outspoken Wookie
2 comments:
Thanks for the infor. Which one is a better workstation for numerically intensive algorithms? I do not need great 3D graphics, and I do not run a server. It seems to be a toss-up.
G'day Peter,
The Core i7 is the desktop processor and the Xeon 5500 is the server processor. So, if you're looking for a desktop system, then an i7 is aimed right at you - the faster the clock speed, the faster it counts.
If you need multiple CPUs, then you'll need to look at an X5500 series CPU as currently, Core i7 CPUs are designed single-processor systems only.
Post a Comment