PDA

View Full Version : New PowerMac G5 and LW 8.2.1 benchmarch!



caesar
04-28-2005, 05:48 AM
Well, I love Macs, but Apple didnt post any details about what scene they used....difficult to replicate the results.
The top model are Dual 2.7 Ghz now, but no dual cores :( Im disapointed. Do you think they need the 65nm chip production to make ´em, or they can use the actual 90nm???
Comes with tiger (OS 10.4)
Man, they got still AGP for the video cards (no PCI-E for now)

Captain Obvious
04-28-2005, 05:52 AM
I don't mean to sound rude, but there are already two threads on this. ;)

caesar
04-28-2005, 05:56 AM
I don't mean to sound rude, but there are already two threads on this. ;)

Sorry, i did a fast check before posting...

Para
04-28-2005, 06:57 AM
Well, I love Macs, but Apple didnt post any details about what scene they used....difficult to replicate the results.

That's a darn shame. Obviously G5 is strong on something but without knowing which scene they've used it's hard to say whether the strength of G5 is in hypervoxels, reflections, radiosity or something completely else.

Limbus
04-28-2005, 09:44 AM
They used skullhead newest but without posting the used settings and the resulting rendertimes this benchmark ist worth anything. Funny thou that every benchmark by Apple, intel or AMD proves that their system is the fastest around.

Verlon
04-28-2005, 10:28 AM
no one is going to try to sell a computer saying "ours is 74% slower in all your favorite applications."

Do YOU shop for the slowest one?

Do you sell your 3D services by saying how much better your competitors are than you? And cheaper?

AMD picks benchmarks that favor them. Apple picks a benchmark that favors them. Intel picks benchmarks that favor them (unless they can't find any, then they buy a benchmark and "re-optimize it" so its faster on Intel and slower on AMD---Bapco anyone?).

Wade
04-28-2005, 11:04 AM
I would like to see a cost per % gained graph. :p

JML
04-28-2005, 11:30 AM
from this thread :
http://vbulletin.newtek.com/showthread.php?t=36301
(same kind of thread)

"for more accurate LW benchmark go there

http://www.blanos.com/Benchmark/

those LightWave benchmark on the mac web site are really weird,
dual G5 2.3 and 2.7ghz faster than a dual xeon 3.6ghz ? right...

on LW benchmark, with the Tracer Radiosity scene,
it takes,
a dual xeon 3.6ghz : 123 secs
dual G5 2.5 ghz : 249 secs
"

JML
04-28-2005, 11:36 AM
from the official Mac web site:
http://www.apple.com/powermac/performance/

".....shredding the PC’s reputation in the process"
"Power Mac G5 is the kind of machine a PC wouldn’t want to compete for a job against"

so much hate against PCs...

harlan
04-28-2005, 11:39 AM
Apple lists the following information on the same page as that benchmark image:

--------------
Benchmark scene = Skull_Head_Newest. For PC systems, cache sizes were: Dell Dimension XPS Gen4 = 2048K L2; Dell Precision 670 = 2048K L2; Alienware Aurora 5500 = 1024K L2; Boxx Tech Series 7300 = 1024K L2.
--------------

In regards to the new G5's lacking dual core functionality, the chips exist and Apple has a plethora of them in testing; the problem is the expense of the technology. Just look at the prices of the Intel/AMD offerings.

PowerMac's are already expensive as it is, and integrating the dual-core chips in them now would only add to that cost. If they were able to implement the dual-core chips into the current G5's while keeping the same price/profit structure, they would no doubt be in them right now.

Rest assured, we've just seen the last update to the current line up of PowerMacs. The next PowerMac revision will more than likely be an entirely new system rather than a simple speedbump, as the most recent speedbump utilizes the maximum the 970 chipset is capable of in its present incarnation.

It's pretty much guaranteed that we'll see Dual-Core's, Faster Chips, and no telling what else in the very near future for the PowerMac. Apple is in the computer business, and they realize that they have to compete in order to be successful. It's just a shame that IBM dealt them a similar blow as did Motorola, but that's the way it goes I guess.

harlan
04-28-2005, 12:01 PM
Funny thou that every benchmark by Apple, intel or AMD proves that their system is the fastest around.


Of course. That's the world of marketing, more commonly known as "deceiving". :)

It's the same concept as NewTek claiming to have an integrated Hair & Fur system in LW; when in all actuality they have a "severely limited, demo version of a 3rd party product".

One of my favorite examples of marketing non-sense is the nVidia website. It's funny how every one of their graphics cards are the most powerful units available. One page will say the 5200 is the "Ulimate in Graphics Processing", and then you look at the 6800 series and suddenly it's the "Ultimate in Graphics Processing"

Take all marketing claims with a grain of salt, as they're meant for nothing more than to make their product look better than everyone elses. I'm also not a big fan of Benchmarking in general, as they're very easily tweaked to "benefit" one product over the other.

My philosophy is to test out the products you're interested in beforehand, and see how well they work specifically for "you". One person may be able to do 10x the amount of work on Commodore 64 as they can on a Dual AMD64 system. Personally, I have no problem with the Windows world at all; I do however prefer Apple systems. My work used to be primarily Windows based until I tried out the Mac; I quickly fell in love with it and realized that regardless of what a benchmark may claim, I'll be able to do more work on the Mac because it works the way "I" think.

It always irritates me to no end, when these fan-boy douchebags start harrassing others about their platform choice. There are quite a few of these people here. It just does no one any good.

james_dmi
04-28-2005, 12:37 PM
Take all marketing claims with a grain of salt, as they're meant for nothing more than to make their product look better than everyone elses. I'm also not a big fan of Benchmarking in general, as they're very easily tweaked to "benefit" one product over the other.

100% agree

It’s utterly pointless to take anything from a benchmark published by a hardware vender. Don’t forget apple have had there hands slapped before for the famous Photoshop filter tests where by they tested all other CPUs using the standard install and then wrote a highly optimised version of the plug-in and ran that on there own hardware. Not saying they did that here but it’s against this background that you should judge. Intel and AMD have done similar things in the past too.

There are independent websites out there that test using Lightwave as a benchmark wait till the machines are available and there are some objective tests and then all will become clear.



My philosophy is to test out the products you're interested in beforehand, and see how well they work specifically for "you". One person may be able to do 10x the amount of work on Commodore 64 as they can on a Dual AMD64 system. Personally, I have no problem with the Windows world at all; I do however prefer Apple systems. My work used to be primarily Windows based until I tried out the Mac; I quickly fell in love with it and realized that regardless of what a benchmark may claim, I'll be able to do more work on the Mac because it works the way "I" think.

100% agree

I made the opposite decision for the same reasons.

JML
04-28-2005, 12:59 PM
Apple lists the following information on the same page as that benchmark image:
Benchmark scene = Skull_Head_Newest. For PC systems, cache sizes were: Dell Dimension XPS Gen4 = 2048K L2; Dell Precision 670 = 2048K L2; Alienware Aurora 5500 = 1024K L2; Boxx Tech Series 7300 = 1024K L2..

this infomation about the cache size is pretty much useless.. this would not affect rendering speed really..
information about how they did the graph will be more useful .
because it's pretty much totally the inverse of the result from Lw benchmark web site.



My philosophy is to test out the products you're interested in beforehand, and see how well they work specifically for "you".
agree



I'm also not a big fan of Benchmarking in general, as they're very easily tweaked to "benefit" one product over the other.

I agree too about most of those benchmarking.. (done by builders themself )
but some of the ones that are done by people themselve are good. (to know the truth)



There are independent websites out there that test using Lightwave as a benchmark wait till the machines are available and there are some objective tests and then all will become clear. .
that's what I meant by :
http://www.blanos.com/Benchmark/

like I said in the other thread, this was not a pc/mac thing but how LW behaved with pc/mac
G5 are good with LW,
it was just funny to see how wrong those LW benchmark were from the apple site.

caesar
04-29-2005, 10:51 AM
In regards to the new G5's lacking dual core functionality, the chips exist and Apple has a plethora of them in testing; the problem is the expense of the technology. Just look at the prices of the Intel/AMD offerings.

PowerMac's are already expensive as it is, and integrating the dual-core chips in them now would only add to that cost. If they were able to implement the dual-core chips into the current G5's while keeping the same price/profit structure, they would no doubt be in them right now.

Rest assured, we've just seen the last update to the current line up of PowerMacs. The next PowerMac revision will more than likely be an entirely new system rather than a simple speedbump, as the most recent speedbump utilizes the maximum the 970 chipset is capable of in its present incarnation.

It's pretty much guaranteed that we'll see Dual-Core's, Faster Chips, and no telling what else in the very near future for the PowerMac. Apple is in the computer business, and they realize that they have to compete in order to be successful. It's just a shame that IBM dealt them a similar blow as did Motorola, but that's the way it goes I guess.

Wow, how do you known that???? But I think the major reason dual-cores G5 are not being released is not price = it´s heat!

G5 has a liquid cooling solution, and its not for the "coolness factor" but for "leak electrity problem" - its a kind of energy lost of about 30%, against 5% of the 130nm producted chips - AMD, Intel and IBM had in the new 90nm chips.

So we need more energy to maintain the performance, but it causes more heat, exactly what made Intel abort any higher GHz in the Pentium chips.

A dual-core processor has exactly the size of an 130nm chip, or twice of a ctual 90nm chip. I think the priceis no so "real state" silicon area cust, but more on "inovation/brand-new" category.
BTW, Intel has an Pentium D for 240 US (ok, you need a new motherboard) running both at 2.8, and AMD Athlon FX 2 will start for 540 US...

Captain Obvious
04-29-2005, 11:40 AM
from the official Mac web site:
http://www.apple.com/powermac/performance/

".....shredding the PC’s reputation in the process"
"Power Mac G5 is the kind of machine a PC wouldn’t want to compete for a job against"

so much hate against PCs...
It's a sales-pitch, nothing else. Everyone claims to have the best products around. Seriously, it's nothing new, nothing special.



PowerMac's are already expensive as it is, and integrating the dual-core chips in them now would only add to that cost. If they were able to implement the dual-core chips into the current G5's while keeping the same price/profit structure, they would no doubt be in them right now.
Why would dual-cores be so much more expensive than dual single-cores? Chances are it's cheaper, in fact.



it was just funny to see how wrong those LW benchmark were from the apple site.
They weren't wrong. I really doubt Apple, or any of the others, for that matter, have outright lied on the benchmarks. They take whatever's the most beneficial, obviously. Why wouldn't they?



Wow, how do you known that???? But I think the major reason dual-cores G5 are not being released is not price = it´s heat!
Nope, it's not heat. The only heat problem with the G5 is heat per area. A dual-core processor is twice as big as a single-core, and thusly there is no heat issue.



A dual-core processor has exactly the size of an 130nm chip, or twice of a ctual 90nm chip.
That depends on how the chip's designed. If they have the same total amount of cache as a single-core version, it won't be twice as big. If they have two separate and larger caches, it will be more than twice as big.

caesar
04-29-2005, 12:11 PM
Nope, it's not heat. The only heat problem with the G5 is heat per area. A dual-core processor is twice as big as a single-core, and thusly there is no heat issue.

I dont know, its clear that bigger processor area will need more cooling, so we have the same situation Intel got = put 2 core only possible at lower clock = their faster Pentium D is a 3.2 only compared to their 3.8 Pentium 4. AMD has a better design, with 2.4 dual core against their Athlon 64 FX with 2.6.
So Apple/IBM maybe needed to lower to 2 GHz each core, and that could be not a greatdeal for them...I just think Why not release a dual-core machine today?



That depends on how the chip's designed. If they have the same total amount of cache as a single-core version, it won't be twice as big. If they have two separate and larger caches, it will be more than twice as big.

Ok, thats example comes from the AMD design, both cores has his 1 MB lvl 2 cache

Captain Obvious
04-29-2005, 03:09 PM
I dont know, its clear that bigger processor area will need more cooling, so we have the same situation Intel got = put 2 core only possible at lower clock = their faster Pentium D is a 3.2 only compared to their 3.8 Pentium 4. AMD has a better design, with 2.4 dual core against their Athlon 64 FX with 2.6.
So Apple/IBM maybe needed to lower to 2 GHz each core, and that could be not a greatdeal for them...I just think Why not release a dual-core machine today?
Keep in mind that a 3.4 or so GHz Prescott P4 will use upwards to 180 watts when you stress it badly enough. A 2.5GHz G5 might reach 80 watts at most, perhaps. The G5's core is much smaller than the P4's, though, so the heat per mm^2 is probably about the same, but the total power usage is so much lower that they shouldn't have real problems getting a dual-core version to run at high speeds. The reason they don't have them out yet is probably a lack of processors. A quad 2GHz PowerMac would be much better than a dual 2.7GHz, for the most part. ;)

Piolla
05-01-2005, 10:03 AM
I like to have all this information, but I'm sure they would be selling this extra super quad dual core no heating ships on the G5 right now if they could do it at a reasonable price. They can't. I can live with that. The one thing I can't understand is why can't I have a professional 3D graphic card on a Mac?
I like to work on Mac OS X, but a render farm with cheaper/faster pcs with linux os whatever os would be the better choice. What bothers me is that I'll not have a good refresh while animating a compicated scene. Or a simple one....
I'm sure it is not a big deal to make drivers for PRO cards to work on Macs. Why can't we have it?

caesar
05-02-2005, 08:50 AM
Truly, I would LOVE to see a dual dual core G5 2.5 Ghz in Tiger = WOAW !!!!

E ai Piolla, sumido da lista (yahoo)??? A Globo "ownz you" hehehhehe?

Ztreem
05-02-2005, 09:45 AM
Quite funny benchmark, it's really not saying anything at all. I mean you really need to see some rendertimes to judge how much faster it is. If the PC renders in 0,117 sec and the Mac at 0,1 sec it's 117% faster but does it really matter. :p