MacWorld post MDD Benchmarks
<a href="http://www.macworld.com/2002/12/features/benchmark.html"; target="_blank">http://www.macworld.com/2002/12/features/benchmark.html</a>;
The dual 1.25GHz looks like it's pretty darn quick. 19% faster than both dual 1GHz machines.
The dual 1.25GHz looks like it's pretty darn quick. 19% faster than both dual 1GHz machines.
Comments
<strong><a href="http://www.macworld.com/2002/12/features/benchmark.html"; target="_blank">http://www.macworld.com/2002/12/features/benchmark.html</a>;
The dual 1.25GHz looks like it's pretty darn quick. 19% faster than both dual 1GHz machines.</strong><hr></blockquote>
I think both the new 1GHz and 1.25GHz machines are a bit of a disappointment in performance terms. There seems to be little difference between the old and new 1GHz machines, despite the new architecture, and the difference between the 1GHz and 1.25GHz machines seems to be no more than what the raw megahertz difference would dictate, despite the larger cache size. <img src="graemlins/hmmm.gif" border="0" alt="[Hmmm]" />
[ 09-18-2002: Message edited by: RodUK ]</p>
I think both the new 1GHz and 1.25GHz machines are a bit of a disappointment in performance terms. There seems to be little difference between the old and new 1GHz machines, despite the new architecture, and the difference between the 1GHz and 1.25GHz machines seems to be no more than what the raw megahertz difference would dictate, despite the larger cache size.</strong>
If Apple improved the core logic performance of the MDD Power Macs, you won't be seeing it in benchmarks because most benchmarks don't measure that sort of performance. What they did was integrate FireWire, Ethernet, and ATA/100 onto the system controller ASIC along with a direct PCI bus. This alleviated bus contention on the PCI bus or older system controller since said items were on the PCI bus or didn't have DMA, nonblocking access, whatever, in the system controller.
One would have to be read or writing across (pushing data across) the network, across the Firewire bus, across the ATA bus, and across the PCI bus while doing something memory intensive to really see if it is better than the QS machines.
As for CPU performance, I would contend that memory technology such as DDR won't provide a large impact on PPC G4 performance as long as there is a large backside cache present. So, performance increases are more directly proportionate to MHz increases than memory performance. On the x86 side, memory performance is more important because x86 CPUs stopped using backside cache (because it was expensive) awhile back, and I get the feeling that the x86 ISA (8 registers) plays a factor as well.
I don't doubt that the 1.25 is faster still - how could it not be. I just think that there's a misconception that the new DP 1 GHz is only as fast as the last one, and I don't believe that's true. With the faster disk bus, faster DDR RAM and faster FSB it makes a noticable difference in my opinion.
If both machines are "stock" however, then the new G4 would win because some of its parts are faster.
Of course, it's a big difference on every Mac, but still.
More benchmarks.
That makes the DDR RAM worth it in my book: Between that and QE, the machines are capable of working under significantly heavier loads than the old machines (esp. running 10.1). I know there's a lot of emphasis on single-app performance, but to me the increased multitasking capabilities are a clear advantage: I rarely do one thing that seriously engages the CPU, but I'm prone to starting things off, backgrounding them if they take a while, working on something else, maybe firing up a game without quitting anything for a few minutes...
It's funny that he writes:
[quote]It's important to note that compared to the old SDR Power Mac, there is still no advantage to having DDR memory. <hr></blockquote>
And them reports seeing a %19 and %22 advantage under load. Extrapolating that data could logically bring you state the conclusion that DDR allows your 1Ghz Mac under load to equal the performance of a 1.25GHZ SDR machine based on guestimating the linear projected speed increase due to the megahertz advantage.
yadda yadda yadda it sounds like DDR will definitely benefit a Multitasker.
Benchmarking all these MDDs without 2GB of RAM in each one is like drag racing with cars that have governors on them set to crap out at 65 MPH. What a waste of human resources.
Are they daft over there at MacWorld or what?
From some benchmarks I've seen the dp 867Mhz machine is a mixed bag compared to the sp 933Mhz.
Curious that Apple didn't make a dp 933Mhz machine as the entry Powermac.
<strong>I'm assuming of course that the dp 867Mhz Powermac is faster than the sp 933Mhz machine only when multitasking, and only with programs that take advantage of the dual processors.
From some benchmarks I've seen the dp 867Mhz machine is a mixed bag compared to the sp 933Mhz.
Curious that Apple didn't make a dp 933Mhz machine as the entry Powermac.</strong><hr></blockquote>
The days of multiprocessor unaware applications are numbered. Moreover, as the system itself is now multiprocessor aware, you are looking at obsolete benchmark data that doesn't reflect the rapidly changing nature of the latest versions of most applications in a multiprocessor aware system context.
The idea that the MDD Macs are slower than previous models is residing in quite a twisted reality distortion field.