Second Ars PPC 970 Article

Posted:
in Future Apple Hardware edited January 2014
Looks like Hannibal has finally finished the second of his two 970 articles. 11,000 VERY interesting words to throw on the performance debate fire. Looks like integer and altivec performance is not as good as expected but the FPU and bus show great promise



Second Ars PPC 970 Article
«1345678

Comments

  • Reply 1 of 143
    shaktaishaktai Posts: 157member
    I understood it a little differently. The integer performance, is indeed not what some folks have hoped for, but about what I expected. It won't blow away any top end P4's on raw integer. However my impression is that the Altivec potential is really more of an unknown with both up and down potentials, but still superior in most aspects to the x86 implementations of SIMD.



    The FPU performance and overall system bus speed does look exceptionally nice and is where the 970 appears to most likely shine. It really emphasizes that there are a number of unknowns still. We will have to wait for real hardware before we know. Overall though, a significant improvement over the present situation, especially if Apple improves the system design as expected.



    In the end, it poses as many questions as it answers. Definitely a lot of food for debate.
  • Reply 2 of 143
    existenceexistence Posts: 991member
    Quote:

    Originally posted by Shaktai

    Definitely a lot of food for debate.



    You can say that again.
  • Reply 3 of 143
    bartobarto Posts: 2,246member
    Can't reply with any thoughts.... Too busy reading most anticipated Mac white paper ever....
  • Reply 4 of 143
    dfilerdfiler Posts: 3,420member
    Quote:

    Originally posted by Barto

    Can't reply with any thoughts.... Too busy reading most anticipated Mac white paper ever....



    I think my head is about to explode and I'm only on page 9 of my second reading. As always, Hannibal offers up an excellent technical analysis... even if I don't agree with all of his extrapolations.



    For instance, his analysis is great for evaluating chip speed in a vacumn. Yet, overall system performance is now more complicated. He seems to use bus,CPU,and memory performance as a benchmark for system performance. However, with QE, GPU performance is now a key factor in desktop speed. Hannibal's analysis is more suited to analyzing server performance for things like render farms, apache, and databases.



    Hannibal has done an excellent job of describing every tree... yet readers should still strive to see the forest.
  • Reply 5 of 143
    bartobarto Posts: 2,246member
    Quote:

    Originally posted by dfiler

    I think my head is about to explode and I'm only on page 9 of my second reading. As always, Hannibal offers up an excellent technical analysis... even if I don't agree with all of his extrapolations.



    For instance, his analysis is great for evaluating chip speed in a vacumn. Yet, overall system performance is now more complicated. He seems to use bus,CPU,and memory performance as a benchmark for system performance. However, with QE, GPU performance is now a key factor in desktop speed. Hannibal's analysis is more suited to analyzing server performance for things like render farms, apache, and databases.



    Hannibal has done an excellent job of describing every tree, but readers should strive to see the forest.




    Which he explains partially in the conclusion.



    Hannible bases what little performance comparisons he does make on a system with DDR400, and no assumptions about cutting edge innovations. This is, after all, a PowerPC 970 article, not a Mac v PC article.



    Barto
  • Reply 6 of 143
    Here's an idea - how about we steal from Ars and make a perpetual 970 thread? That way we won't end up with 5,000 970 threads sprouting every time something new happens. It makes it very hard to read what's new in the rumor world.



    (In other words, "In before lock!" )
  • Reply 7 of 143
    netromacnetromac Posts: 863member
    Quote:

    Originally posted by Anonymous Karma

    (In other words, "In before lock!" )



    you just got yourself banned there Anonymous Karma
  • Reply 8 of 143
    programmerprogrammer Posts: 3,457member
    Quote:

    Originally posted by jamm

    Looks like Hannibal has finally finished the second of his two 970 articles. 11,000 VERY interesting words to throw on the performance debate fire. Looks like integer and altivec performance is not as good as expected but the FPU and bus show great promise



    Second Ars PPC 970 Article




    Its an interesting article, but remember it is heavily based on opinion and supposition. There are big unknowns in the article and he fully admits not having that information, and that that information could dramatically effect performance (in either direction).
  • Reply 9 of 143
    outsideroutsider Posts: 6,008member
    Especially regarding L1 & L2 cache latency and VMX performance in which he knows nothing about apart from the very basics.
  • Reply 10 of 143
    producerproducer Posts: 283member
    What's nice is this chip should scale well.. if the rumors are true 2.5 by the end of the year....looks like it might even scale better than the pentium...
  • Reply 11 of 143
    hmurchisonhmurchison Posts: 12,419member
    I tend to think that performance will be better than stated. I truly hope Apple comes up with a 2Ghz Model on launch. That would help greatly.



    Also we're not taking into account Panther as well. Apples timing couldn't be better. We have the potential of new Hardware running much faster and yet a 3rd Major update to the OS coming as well. I think the Real World speeds of PPC 970 Macs will be equal to and superior to Wintel. We must stay within 2/3 of the P4 but OSX will give us a 1/3 advantage meaning a Dead Heat on average. And that bodes well for Apple.
  • Reply 12 of 143
    shaktaishaktai Posts: 157member
    Quote:

    Originally posted by Programmer

    Its an interesting article, but remember it is heavily based on opinion and supposition. There are big unknowns in the article and he fully admits not having that information, and that that information could dramatically effect performance (in either direction).



    Absolutely. Then all articles of this nature are going to be heavily based on opinion (the authors) and varying degrees of suppostion. The article provides a lot of information and obviously a lot of work was put into it. In the end though, until we get actual production 970's to examine and test, and see what kind of a system Apple builds around it, there will be no real answers. The technical analysis seems good, to the extent that the information was available for analysis. Still a lot of missing pieces. From here on out until the final product is released, all we have is opinion, suppostion and rumors. I doubt that we will see any more verifiable information before introduction.



    From my perspective, the article provided insights into what types of operations we "may see" the new systems excell at and what types of operations they "will likely" lag behind at.



    It looks promising for my needs. The areas where it excells, may be significant, and the areas where it lags will be minor. Still a huge improvement across the board from where we are now. 8)
  • Reply 13 of 143
    lemon bon bonlemon bon bon Posts: 2,383member
    Quote:

    As with all things Apple, though, the big question is price. Will Apple drop its margins drastically and sell these machines at a competitive price point in order to increase market share, or will it continue to price itself into the increasingly non-existent luxury/lifestyle computing niche?



    That IS the question.



    Apple have a real chance at this point in history to slay their last sacred cow. It won't come their way again for a long time.



    Lemon Bon Bon
  • Reply 14 of 143
    shaktaishaktai Posts: 157member
    The Titan City site has also posted a preliminary analysis of the 970 as it would be used in a Mac. Short, but in simple enough language as to make some sense to non-techies. Can't vouch for the accuracy of their assumptions, but it made sense to me, and the article is in English. http://haxor.dk/index.html
  • Reply 15 of 143
    hmurchisonhmurchison Posts: 12,419member
    Apple issues with Price are a fallacy



    A brand new Mac with a Monitor and G4 can be had for $799



    a Top of the Line unit can be had for under $3000.





    What I believe people are saying is that they wish the unit of "their" desire was cheaper. I contend that Mac users complaints on Price only extend as far as their own value proposition. Those that complain about the price almost always point to the Powermac line and point to the lineup as being underpowered. So...in essence they aren't complaining about the Price but what they are "getting" for that price.



    So really I think you'll find that Apple doesn't have to lower prices. If the PPC 970 system comes and the performance is what we expect then these units will sell by the bushell. People want to know that when they purchase a computer ..it's fast and will last a while. Price of course is a consideraton but for some not as important as others.
  • Reply 16 of 143
    rickagrickag Posts: 1,626member
    Quote:

    Originally posted by Lemon Bon Bon

    That IS the question.



    ..........It won't come their way again for a long time.



    Lemon Bon Bon




    Oh, I don't know. There's the rumoured IBM 980.



    Logically, the next generation offspring of the 75X, the 85X ??, that may marry the 440 and 75x technologies(re: Multicore Superscalar, Rapid I/O, n-way Coreconect, VMX, etc.), and may see the light of day in a NUMA architecture.



    Or maybe some version of IBM's cell technology.



    But it sure would be nice if the price for any new IBM 970 based computers from Apple stay at the current price points or even drops somewhat. Will Apple finally try to capture more market share or not, hmmm? We'll see in the next couple of months. My money is now stashed after years of pack ratting and I'm just waiting.
  • Reply 17 of 143
    yevgenyyevgeny Posts: 1,148member
    Something very important to bear in mind when comparing the performance of these machines to Intel's current offerings is that the high end machines will be dual CPU machines. Where one 970 chip will compare quite favorably to a single P4, two 970's will completely beat any P4 when running MP aware software. Valid comparisons will require a dual P4 machine (not quite a common in the windows world as in the Mac world).



    These machines are going to be incredible.
  • Reply 18 of 143
    amorphamorph Posts: 7,112member
    There's a fair amount of talk in that article and at Ars about "whether" Apple will slash their profit margins for market share.



    This is not a rumor, or a possibility. or unlikely. Fred Anderson's been banging this drum for over a year now. Apple has to move carefully, because of the economy and because they're breaking even operationally, but they just launched what looks like a solid new revenue stream, and every bit helps.



    Also, hmurchison is right in that you can improve price/performance by dropping price or by increasing performance (or both). Apple has a couple of ways to bump performance (the next version of OS X and the 970, and maybe some other goodies), so even if they hold the line on price their equipment will be that much more desirable. Of course, if they drop prices further, Macs will look even better.



    I think the next few months will be remarkably interesting, and refreshing after three years of the G4.
  • Reply 19 of 143
    yevgenyyevgeny Posts: 1,148member
    Quote:

    Originally posted by Amorph

    This is not a rumor, or a possibility. or unlikely. Fred Anderson's been banging this drum for over a year now. Apple has to move carefully, because of the economy and because they're breaking even operationally, but they just launched what looks like a solid new revenue stream, and every bit helps.





    Apple is breaking even with virtually nonexistent Pro sales. If you aren't selling G4's, then you aren't making any money on them. Therefore, Apple can afford to lower margins on their pro machines because where they will make less money per box, they will actually be selling boxes. Apple should go for lower margins on everything but the highest end 970.
  • Reply 20 of 143
    jammjamm Posts: 37member
    Here is how I see Apple's future in terms of performance and new technologies:
    • The introduction of the 970 will usher in a new mac era, capable of performing on a par with or better than Intel/AMD offerings based on real-world application perfomance

    • This will only be achieved if we see a rapid spread of the 970 (and perhaps later the 980) to consumer/portable machines. In this respect, Apple should move to a single processor platform - drop the G4 ASAP. I guess this means putting all your eggs in one basket but Apple has a knack of pulling this off (68K - PPC, OS 9 - OS X...). However, the cost savings of buying in bulk and only having to support one motherboard and processor architecture can only be beneficial.

    • The differentiation between consumer and professional machines will be severely blurred by following this route - Apple will have to solve this problem as it has already arisen with the iMacs getting a G4. Perhaps MP, >cache, new interconnect technologies (FW800, PCI-X)...

    • Those Mhz ratings must go UP UP UP if we want market share to grow! The Mhz wars are unknown to the general public. Show Joe Six-Pack a 2 GHz+ Intel and a 1 Ghz iMac. Sure, he may think the iMac looks all 'pretty' but what hes thinking is that 2 is better than 1 and hence he is getting more bang for his buck. Apple has failed time and again in trying to show that 'Mhz doesn't matter'.

    I will now go smoke a J ....
Sign In or Register to comment.