Early Benchmarks: Dual 1GHz

[arn]

Many Mac news outlets and readers have pointed out Barefeats' early benchmarks of the Dual 1 GHz G4 (DDR) vs Dual 1 GHz G4 (non-DDR).

Initial results appears to be identical... however, base configurations of the Dual 1GHz G4 (non-DDR) has 2MB of L3 cache vs 1MB of L3 cache on the DDR G4.

As well, others argue that the tests don't actually stress the particular systems enough to measure differences in the bus speeds. (Dennis of http://www.fractaldomains.com/)

 

[chewbaccapits]

 

[MikeH]

Having taken a quick look at the test results I would have to agree with those who say that the tests do not stress the machines.

Something that renders in Bryce in less than a minute is hardly putting the hardware through its paces is it.

Give 'em some work to do; Radials blurs on 60mb images, complex global illumination renders - stuff that take 30 minutes or so, not seconds.

Then lets moan if there's no improvement.

 

[Sun Baked]

Looking at the 133 MHz bus diagram for the DDR tower that Arcady kindly left here
https://forums.macrumors.com/showthread.php?s=&postid=130228#post130228

and marking up the diagram a little, you see that on a 133 MHz bus, even if you give the MaxBus everything it wants, you still have memory bandwidth left over for video and storage. And the same would still be the case on the 166 MHz bus.

So why on all these 3 multitasking/SMP machines should all the benchmarks be so similar?

Especially when 2 are DDR, 2 have 133 MHz buses, 2 have 2 MB/CPU of L3 cache, 2 have ATA-100, and 2 have G4 Ti video

 

[deepkid]

Again, too much left to question...

As we've wondered in another thread..

1. What OS and version was used? Jaguar would probably be the best gauge.

2. How much RAM was in each machine?

3. What type of hard drives and what were their speeds? Were these identical?

Also, in the diagram it says that these are Apollo (7455) chips?? Not 7470, huh?
Notice that on the roadmap below..the Apollo chips weren't expected to scale past 1 Ghz.

So 7455 chips would sorta make sense in th low and mid-range machines.

However...

It should make you wonder if Apple will ship the top-end, dual-1.25 powermacs with a different chip.. maybe the 7470, which would make things very interesting. According to the roadmap, those are projected to reach 1.25 Ghz.

I don't know how close real world chip output compares with this roadmap, but maybe there's still a chance that the top end machine might be a different beast all together. (Unless someone can factually say that Apple won't be using anything different.)

http://www.geek.com/procspec/apple/g4.htm

 

[soilchmst]

Re: Again, too much left to question...

Originally posted by deepkid
As we've wondered in another thread..

1. What OS and version was used? Jaguar would probably be the best gauge.

2. How much RAM was in each machine?

3. What type of hard drives and what were their speeds? Were these identical?[/url]

This test seems highly spuspect with no listing of the equipment in the computers tested. Maybe they only tested a standard model with the slower Radeon 9000 vs. the old dual 1 Ghz Geforce 4 Ti.

Also, I noticed on Mac centrals link to the tests that a reader ran these tests. Is that Barefeet's normal practice?

 

[ryan]

While I guess it is nice to see that the current middle-of-the-line PM G4 is as fast as the former top-of-the-line PM G4 it seems to me that there shouldn't have to be tests designed specifically to stress the machine to show how the (hacked) implementation of DDR is of benefit to the new machines. The new machines should simply be faster, regardless of what they're doing, than the machines they're replacing, period.

 

[ImAlwaysRight]

It seems to me this is another case of trying to boost number of hits at a site (barefeats.com). These 4 tests are all CPU tests, and do not stress the system. Being CPU tests, the L3 cache will make a difference. Where are 2D scroll tests? Where are 3D game tests? These are part of his normal testing as given in his "how I test", yet he either chose not to run them, or ran them and didn't like the faster results so didn't report them. It seems to me the "reviewer" is biased and probably upset we didn't see 1.2/1.4/1.6 G4's so is going to get back at Apple. OK, I'm reading a lot into it, but publishing results like this are LAME. Also, if you've been around a while, you know that barefeats results have been suspect in the past.

The only accurate test would be to have the same amount of RAM and move the hard drive and video card from one machine to the other, or AT LEAST report what the differences are. We don't have any of this here, so I'd ignore these tests for the time being.

But if you choose not to, then also relish the fact that a dual 867MHz is 45% faster than a 2.54GHz Pentium 4 computer.

 

[oranjdisc]

Truth is, I'm actually kind of glad to see Bare Feat's results popping up on all these news sites. Apple lowered the PowerMac's prices, but they purposely crippled the machines to maintain their bloated profit margins.

 

[ImAlwaysRight]

Originally posted by ryan
While I guess it is nice to see that the current middle-of-the-line PM G4 is as fast as the former top-of-the-line PM G4 it seems to me that there shouldn't have to be tests designed specifically to stress the machine to show how the (hacked) implementation of DDR is of benefit to the new machines. The new machines should simply be faster, regardless of what they're doing, than the machines they're replacing, period.

Ryan, not necessarily so. If you have a CPU bound task, then putting a dual 1GHz vs. a dual 1GHz isn't going to change much. The only difference here is old gig has 2MB L3 cache, and new gig has 167MHz bus, so based on these 4 CPU tests, these two seem about a wash with the slight edge going to L3 cache.

But, when you consider the other improvements to the new dual G4, like case design (dual CD/audio in/front headphone/4 RAM slots/3 ATA buses/etc), and the fact that in other tests like games and 2D scrolling the new stock G4 dual 1GHz will beat the old dual 1GHz, and comes with Jaguar, and costs $500 less, I think that is a pretty good upgrade. However, considering the old dual gig is now $2199, that makes it a pretty good buy, considering more RAM and 2MB L3 cache/processor. That is a tough choice for prospective buyers. But, if you want an old gual gig, you better act fast, as once they are gone, they are gone.

I'm eligible for Apple edu discount, so picked up a new dual gig for $2249 shipped (no tax). After paying shipping at a mail order house, it is about the same price as the old dual gig, and I like the newer better.

 

[Timo_Existencia]

Ridiculous responses...

The denigration of Bare Feats is getting ridiculous. He has posted some results that I find interesting and of themselves, dissapointing. Clearly, it is early, and he doesn't have a new box in his own hands. But to ascribe some conspiracy theory to him is a bit over-the-top.

I agree with the earlier poster who stated that you shouldn't need to craft a perfect task in order to find the manner in which this new machine is faster; according to the Apple Hype, this machine should simply be faster across the board. That it is not is troubling.

I always become worried when Apple Zealots turn into apologists; I thought I had left the dogmatic religion of my childhood, but wonder sometimes If I didn't merely replace it with my devotion to Apple. Reading some of the comments here and elsewhere make me believe that dogmatism is alive and well in the Apple community.

 

[ibjoshua]

Originally posted by oranjdisc
Truth is, I'm actually kind of glad to see Bare Feat's results popping up on all these news sites. Apple lowered the PowerMac's prices, but they purposely crippled the machines to maintain their bloated profit margins.

huh?

why would you 'cripple' a machine to maintain a profit margin. using cheap parts maybe, but actually 'crippling'? do you understand the meaning of the word?

j

 

[Timo_Existencia]

Where?

However, considering the old dual gig is now $2199, that makes it a pretty good buy, considering more RAM and 2MB L3 cache/processor.

Where is the old 1 gig being sold?

 

[ImAlwaysRight]

Re: Ridiculous responses...

Originally posted by Timothy
The denigration of Bare Feats is getting ridiculous. He has posted some results that I find interesting and of themselves, dissapointing. Clearly, it is early, and he doesn't have a new box in his own hands. But to ascribe some conspiracy theory to him is a bit over-the-top.

I agree with the earlier poster who stated that you shouldn't need to craft a perfect task in order to find the manner in which this new machine is faster; according to the Apple Hype, this machine should simply be faster across the board. That it is not is troubling.

Folks, stop and think about this a minute. You have two dual gig Powermacs, the main difference being the old has 2MB L3 cache per processor, the new has 1MB L3 cache per processor. The old has 133MHz system bus, the new 167MHz system bus. Why do you expect much of a performance difference between the two?

Also, some seem to be forgetting the new dual gig IS THE MID-RANGE POWERMAC. As such, compare it to the former mid-range PowerMac, a single 933Mhz G4 w/133MHz bus. The new mid-range will tear up the old. Why is everyone so anal on comparing the new dual gig vs. the old? When in Apple's history has so much grief been given over the new mid-range Powermac? The thing got a speed bump, dual processors, many other improvements with the redesigned PowerMacs, and yet all this disappointment. I understand the 1.25 isn't shipping yet, but when it does, then compare the old dual gig to the new 1.25 which HAS 2MB L3 cache per processor and then we'll see what's going on.

Why are people complaining so? Some have come out and said they are disappointed we didn't get 1.2/1.4/1.6GHz processors. But 1.25 is consistent with what Apple has done in the past. Get over it already. Don't bash the new Macs because of your perception that Apple has short-changed us. It is a good upgrade consistent with what Apple has done in the past.

And Timothy, my comments have more to do with common sense than dogmatism. I know you're still fuming over the "failure of Apple to produce" because of the 1.2/1.4/1.6 G5 rumors of several MacWorlds ago, but get over it already.

 

[ImAlwaysRight]

Originally posted by i_b_joshua


huh?

why would you 'cripple' a machine to maintain a profit margin. using cheap parts maybe, but actually 'crippling'? do you understand the meaning of the word?

j

Exactly. I want to make this point very clear, since many might have missed it in my last post. THE NEW DUAL 1GHz IS THE MID-LEVEL POWERMAC. Repeat after me. THE NEW DUAL 1GHz IS THE MID-LEVEL POWERMAC. Compare it to the Quicksilver 933MHz with one processor and 133MHz bus. Because this is the upgrade we got. PLEASE NOTE THE 1.25GHz with 2MB L3 cache per processor is the new high end model.

Now maybe some won't feel so cheated.

You shouldn't expect a huge perfomance difference comparing two dual 1GHz models against each other. Sure, too bad Apple didn't put 2MB L3 cache in the mid-level Mac, as then it would have been faster with the faster bus, but the new mid level is a LOT faster than the old mid-level. I'm glad I waited to purchase until now.

For those who want to see another test, (again, from Apple's propaganda), also note a new dual 867 beats a 2.54GHz Pentium 4 in DVD decoding.

Geez, since two tests results from Apple's website I've posted to this thread alone show the dual 867 G4 to be faster than a single 2.54 Pentium 4, I wonder if all Pentium 4 owners are going to jump ship and run out and buy the dual 867Mhz PowerMac which not only "performs better" but costs less, too. Shouldn't the Pentium "simply be faster across the board?"

 

[nero007]

Re: Re: Ridiculous responses...

Originally posted by ImAlwaysRight
Folks, stop and think about this a minute. You have two dual gig Powermacs, the main difference being the old has 2MB L3 cache per processor, the new has 1MB L3 cache per processor. The old has 133MHz system bus, the new 167MHz system bus. Why do you expect much of a performance difference between the two?

Also, some seem to be forgetting the new dual gig IS THE MID-RANGE POWERMAC. As such, compare it to the former mid-range PowerMac, a single 933Mhz G4 w/133MHz bus. The new mid-range will tear up the old. Why is everyone so anal on comparing the new dual gig vs. the old? When in Apple's history has so much grief been given over the new mid-range Powermac? The thing got a speed bump, dual processors, many other improvements with the redesigned PowerMacs, and yet all this disappointment. I understand the 1.25 isn't shipping yet, but when it does, then compare the old dual gig to the new 1.25 which HAS 2MB L3 cache per processor and then we'll see what's going on.

Why are people complaining so? Some have come out and said they are disappointed we didn't get 1.2/1.4/1.6GHz processors. But 1.25 is consistent with what Apple has done in the past. Get over it already. Don't bash the new Macs because of your perception that Apple has short-changed us. It is a good upgrade consistent with what Apple has done in the past.

And Timothy, my comments have more to do with common sense than dogmatism. I know you're still fuming over the "failure of Apple to produce" because of the 1.2/1.4/1.6 G5 rumors of several MacWorlds ago, but get over it already.

Lol. I'm with you on this one.

 

[colewave]

Well, there definitely is room for complaining here

Testing the old dual 1ghz against the new dual 1ghz is the best way to judge the other changes in architecture Apple is touting. If the results are "no difference", then you might as well find an old dual 1ghz on deep discount instead of paying up for the new architecture. I agree that the barefeats test doesn't prove anything yet, so I'll wait and see.

 

[ImAlwaysRight]

Re: Well, there definitely is room for complainging here

Originally posted by colewave
Testing the old dual 1ghz against the new dual 1ghz is the best way to judge the other changes in architecture Apple is touting. If the results are "no difference", then you might as well find an old dual 1ghz on deep discount instead of paying up for the new architecture. I agree that the barefeats test doesn't prove anything yet, so I'll wait and see.

(I can't seem to get off this topic for some reason...)

I suppose I'm just a fan of accurate, responsible testing. Let's see the whole picture. If you purport to be objectively testing two systems against each other, then do it. It just doesn't seem like this is what is going on at Barefeats.com.

If you want to judge the changes in architecture then you need to move the same hard drive and video card from one machine to the next. Have the same amount of RAM installed. This is the only way to give an accurate result. Barefeats.com did not do this.

There is some merit in testing the stock configurations of new and previous generations against each other. How much better is the new video card? What difference does the reduction of L3 cache make? But these kind of comparisons should only be done between the same class level, like low-end vs. low-end, and high-end vs. high-end.

It seems the point between comparing the former high end dual 1GHz vs. the now mid-level dual 1GHz IS architecture. So do it right. That's all I'm saying.

I'd be very interested to see tests showing the performance difference between new and old generation PowerMacs IN THEIR OWN CLASS LEVEL. Put the following against each other:

Stock Quicksilver 800MHz vs. new dual 867Mhz
Stock QS 933 vs. new dual 1GHz
Stock QS dual 1GHz vs. new dual 1.25 GHz

EDIT: The largest performance increase would come in the low end system. You have a combination of faster MHz, addition of a second processor, AND additon of L3 cache. The mid-level is a big improvement with faster MHz, faster system bus (166 vs. 133), and addition of second processor, but the 933 had 2MB L3 cache while the dual 1GHz has 1MB L3 cache, although taking all upgrades into consideration, the new will still pummel the old. The new dual 1.25 has the biggest speed jump, 250MHz per processor. It also benefits from faster 166MHz bus. Not to mention all new systems sport ATA/100 and the DDR RAM, which, it seems, doesn't add a whole lot to system performance, but should offer a slight improvement with some tasks. LOOKS TO ME LIKE THESE ARE GOOD UPGRADES.

 

[porovaara]

These benchmarks.

I'm suprised at the ignorance being shown here and touted as the gospel. The simple fact is barefeats does pathetic benchmarks. They have been shown in the past to be of poor methology and with the tasks chosen to "give" a result the tester wanted.

In real world interactive usage the new dual gig blows away the old dual gig. On things like encoding video, making mp3s, and other cpu bound intensive tasks the performance is almost identical. This is amazing considering the g4 really needs that extra L3 cache... amazing because the new dual gig has half the cache and manages to keep up.

Compared to last week the apple world is a lot better. Still not great but the entry level machine is awesome and the dual gig for the midrange is pretty sweet as well as long as you stay in the apple world.

 

[AmbitiousLemon]

i think some of you need to actually LOOK at barefeats.com before attacking the benchmarks. you are screaming and yelling about things that are flat out wrong. he does say very specifically how each machine is configured and he also says very specifically how he set up the tests. as for claiming these are cooked. BS. they dont say what you like so you attack it. claiming he would do this on purpose is completely wrong. barefeats.com is in no need of pulling a stunt to get hits. his benchmarks in the past have been very reliable, and have been very fair.

comparing all the dual 1ghz machines is the best way to test whether the new bus and ddr improve performance. if you can not understand basic scientific method then please dont criticize (putting your response in bold does nothing to convince others yo are right). apple has been telling us that the new dual 1ghz is faster than the previous by between 20 and 50% (in various tasks), thats why he compared it. Also Apple has even claimed the new dual 867 is faster than the previous dual 1ghz. Apple says this is because of the new architecture. Also i am sure as soon as the dual 1.25ghz ships he will add it, he is working with what he can get.

btw more tests ARE on the way aimed at proping the new architecture even more are on the way.

 

[yadmonkey]

Re: Re: Ridiculous responses...

Originally posted by ImAlwaysRight
Folks, stop and think about this a minute. You have two dual gig Powermacs, the main difference being the old has 2MB L3 cache per processor, the new has 1MB L3 cache per processor. The old has 133MHz system bus, the new 167MHz system bus. Why do you expect much of a performance difference between the two?

Also, some seem to be forgetting the new dual gig IS THE MID-RANGE POWERMAC. As such, compare it to the former mid-range PowerMac, a single 933Mhz G4 w/133MHz bus. The new mid-range will tear up the old. Why is everyone so anal on comparing the new dual gig vs. the old? When in Apple's history has so much grief been given over the new mid-range Powermac? The thing got a speed bump, dual processors, many other improvements with the redesigned PowerMacs, and yet all this disappointment. I understand the 1.25 isn't shipping yet, but when it does, then compare the old dual gig to the new 1.25 which HAS 2MB L3 cache per processor and then we'll see what's going on.

Why are people complaining so? Some have come out and said they are disappointed we didn't get 1.2/1.4/1.6GHz processors. But 1.25 is consistent with what Apple has done in the past. Get over it already. Don't bash the new Macs because of your perception that Apple has short-changed us. It is a good upgrade consistent with what Apple has done in the past.

And Timothy, my comments have more to do with common sense than dogmatism. I know you're still fuming over the "failure of Apple to produce" because of the 1.2/1.4/1.6 G5 rumors of several MacWorlds ago, but get over it already.

Good show ol' bean! I agree that these benchmarks are questionable at best. But even the believers seem to be missing out on two points, one of which you mentioned and one of which you hinted at:

1) The new "Faster" (dual-1ghz) Mac is much faster than the old "Faster" (single 933mhz) Mac. Likewise, the new "Faster" Mac has about the same performance as the old "Fastest" (dual 1ghz) Mac. In other words, if you want the previous top performer, it is much cheaper now. Everything is about as it should be.

2) If you believe in these benchmarks, they do indicate one important thing: That the new "Faster" (dual-1ghz) Mac does seem to get some performance boost from the new faster bus and DDR RAM. Why? Well, despite being crippled by a smaller L3 cache than the previous "Fastest" (dual 1-ghz) Mac, it performs about the same. Something is obviously making up for the half-sized L3 cache!!!

It stands to reason that the new "Fastest" Mac (dual-1.25 ghz) will see quite a performance boost over the others, since it combines the 2MB L3 cache with the DDR RAM and faster bus. The moral: If you want to have your cake and eat it too, you must pay Apple for the new "Fastest" Mac. Now who can argue with that?

 

[ElRayOX]

Re: Again, too much left to question...

Originally posted by deepkid
As we've wondered in another thread..

but maybe there's still a chance that the top end machine might be a different beast all together. (Unless someone can factually say that Apple won't be using anything different.)

Deepkid makes a real good point. Maybe the September machines will be completely different beasts...

 

[Pin-Fisher]

Originally posted by ImAlwaysRight
I wonder if all Pentium 4 owners are going to jump ship and run out and buy the dual 867Mhz PowerMac which not only "performs better" but costs less, too. Shouldn't the Pentium "simply be faster across the board?"

Costs less??? How to you figure that the dual 867 costs less? Give me your breakdown of costs between the two system. I'm not talking complete Pentium system vs bare G4 system, Im talking soup to nuts. Monitor , speakers (why the heck Apple doesn't include the speakers Ill never know) If I had a nickle for every time I have heard someone at CompUsa say "Speakers are extra??" then walk over to the InSmell computers I could buy a Porsche.

 

[yadmonkey]

One more thing

A 25% (at least) performance increase on the top model is damn good! So is an $800 price drop on the previous top model. I work in a multi-platform environment and I have multi-platforms (2 macs and 2 PCs) at home, so I have no reason to be partial to a platform other than preference. I can honestly say that these new Dells we got with the really fast bus (533 mhz, I think) and super-over-clocked P4s are overcompensating for something. Why do you use a Mac?

Get you mind out of the numbers game and just look at what is out there - a dual-1ghz Powermac, last-gen or this-gen is a great value for the $$, as is the dual 867. If you can afford the dual-1.25ghz powermac, all the better.

Ask yourself what you need and what you are willing to spend. When you consider the new models and the sales on the previous ones, you should come out with a great deal. I can't remember a better time to pick up a Mac.

 

[deepkid]

Re: Ridiculous responses...

Originally posted by Timothy

I agree with the earlier poster who stated that you shouldn't need to craft a perfect task in order to find the manner in which this new machine is faster; according to the Apple Hype, this machine should simply be faster across the board. That it is not is troubling.

That is a very troubling statement. It is as scary as some court of law simply glancing at an accused person who "might" look guilty and charging them so without due process. Isn't that a bit like witch hunting?

If the tests that Barefeats performed are suspect, they should be not be held in such high regard. It does not matter what Apple's or Barefeats' opinions are, the results should be based upon sound benchmarking.