Apple throws out the rulebook for its unique next-gen Mac Pro

1606162636466»

Comments

  • Reply 1301 of 1320
    drblankdrblank Posts: 3,383member
    Quote:

    Originally Posted by wizard69 View Post





    I don't really know if Apple is a member or not. They don't always publicly acknowledge memberships in such organizations. In any event if they aren't a member they should be

    It makes no sense to argue about something if you are right. The problem is when you are wrong.



    In this regard OpenCL isn't the long term play to support many languages. One example is Java which initially can sit on top of OpenCL however their goal is to have Java working directly with HSAIL by 2015. I believe that is Java 9. I'd actually be surprised if they deliver by 2015.



    The point is HSAIL is a complement to and works in conjunction with OpenCL.

    I've been following this stuff for some time now. There is still a lot of infrastructure that needs to be moved into place before all of the hardware advantages of HSA are realizable. This is my primary point.

    Neither am I but it is helpful to understand where things are going.

    Apple will likely go their own way no matter what the rest of the world does. That is neither bad nor good. It would be nice to see Apple support a common industry infrastructure though. Will they do so? I have no idea.

    Well like I said there is no certainty that they aren't involved. Especially with Apple whom have been known to keep their memberships in various organizations quiet. As for Intel they seem to be flopping around a bit, I doubt they are involved but they really should be. As it was Intel had to be dragged into the OpenCL world and GPU compute in the first place.



    Will the consortium make a difference, I'd have to say yes even if Apple isn't involved directly. Apple will have to keep an eye on development there to make sure what the HSA foundation has to offer doesn't eclipse what Apple offers. By the time the platform matures (probably 2016) Apple may change their minds or at least offer an alternative solution. If the Java and support for other languages, works out in the way the HSAFoundation is projecting it could become a very compelling platform for developers. Apple might see that as a threat.

    I think you are playing mental masturbation and I'm not interested in playing that game with you. You are on your own.

     

    Please stop replying as I will NOT read any of what you are saying because you aren't saying anything that I think is important enough or relevant enough to discuss it.  Some of your posts don't even make any sense or logic to it.  Seriously, don't worry about it.  You'll probably be much better off not to know.  Let's the engineers that are paid to figure things out.

  • Reply 1302 of 1320
    wizard69wizard69 Posts: 12,755member
    drblank wrote: »
    Here's the problem I see with some people thinking that a tablet is going to fully replace a laptop or desktop? For certain things it does, but for other things it doesn't.  Would more RAM be useful in a tablet? Sure it would, but unfortunately, tablets and smartphones have space limitations, power limitations, price/profitability concerns, etc. and until Apple can implement more RAM, they are kind of limited, so they have to keep the OS as lean as possible.
    I don't see tablets replacing all of computing devices either. I do see tablets as a compliment to desktops. In fact for home use I'm really tempted to buy a desktop when it comes time for my next upgrade.

    I personally think that next year, Apple will have more on board RAM in their upcoming tablets, how much more?  I couldn't tell you.  Most likely 2GB, any more than that I don't know.  Believe it or not, there are still plenty of people using the iPad 2, which has only 512MB of RAM and for what they use it for, it's enough.  For some apps, it's best to run them by themselves with no other apps running, but a lot of apps can run just fine with other small apps running.  But it's pretty easy to stop an app from running.
    I was hoping for more this year. Mind you not every app needs more RAM on an iPad but the key app for me, Safari, does need access to more RAM. I probably shouldn't complain because Apple has made some amazing improvements in a very short time period.
    But I don't think that a tablet is the only computing device I'm ever going to use.  And for me and a LOT of others, I think it's kind of silly to ditch everything just to use a tablet.   What Apple uses for RAM in future models is most likely going to be some form of DDR memory and that's about all I need to really know other than how much and if I can afford or need to upgrade what I'm currently using for a new model.  Same rules apply to the MacPro.
    What Apple actually uses in the next machines isn't as important as getting an increase in available RAM. Memory Cube is interesting in that it defines a standard component that uses a stack of chips which increases storage density. The fact that it is relatively low power and very fast is simply icing on the cake.
    Personally, you are getting WAY off topic, which in turn makes everyone else get off topic discussing what you are talking about.
    You are right there. I just see the tech solving issues for Apple. RAM is literally a bottle neck for today's processors. Apple vastly improved the disk bottle neck by going to PCI Express based storage and many argued that that would not happen. Going DDR 4 would certainly be an incremental increase in performance but is that enough?
    So, can we just end this display of mental masturbation and get back to discussing the MacPro that's going to be shipping next week? That's more of an immediate thing that's happening and it uses DDR3 memory, which is NOT going to change for this year's models.
    Actually I'd love to see a few posts from actual owners of the new machine. I know I won't be an owner this week and probably not next year. However once the Mac Pro arrives it is no longer future hardware so the discussion would go in one of the other sections. Strange as it might be future hardware is a discussion forum about things that don't exist.
  • Reply 1303 of 1320
    frank777frank777 Posts: 5,786member
    Quote:

    Originally Posted by drblank View Post

     

    Well, I don't know of MacLife really understands how TB and TB 2 are still better for USB 3.1.  For Joe Blow consumer, they are usually into whatever is the cheapest, but pros are usually interested in what's the best.  TB and TB2 are better than USB3.1.  So, go enjoy your cheap solution late next year, because that's about the soonest you'll see USB 3.1 products coming out.


     

    Yeah, but PCIe outside the box will offer the speed of Thunderbolt without most of the cost. And it's out mid-2014.

     

    Given that TB2 is just out and it can take a year to develop and market a product, will third-parties support it?

     

    Will TB2 co-exist with PCI OTB?

  • Reply 1304 of 1320
    wizard69wizard69 Posts: 12,755member
    drblank wrote: »
    I think you are playing mental masturbation and I'm not interested in playing that game with you. You are on your own.
    I don't think you know what you are talking about here. Sadly much of the post I responded to was questions that you asked. If you don't want an opinion or facts why ask the questions.
    Please stop replying as I will NOT read any of what you are saying because you aren't saying anything that I think is important enough or relevant enough to discuss it.  
    Too bad for you. If you are going to ask questions and then reject the answers who here has a problem.
    Some of your posts don't even make any sense or logic to it.  Seriously, don't worry about it.  You'll probably be much better off not to know.  Let's the engineers that are paid to figure things out.
    OK if you say so. Honestly though if I don't make sense then there are an awful lot of people involved in hardware and software development that don't make sense. The goals that the HSAFoundation has come right from their web site. It might not make sense but that is only because there is a lot of confused posting going on out there.
  • Reply 1305 of 1320
    wizard69wizard69 Posts: 12,755member
    frank777 wrote: »
    Thought I'd catch up on some reading tonight, I saw that MacLife's January issue has an epic takedown of Thunderbolt.
    Haven't seen that, mainly because I can't stomach MacLifes style and lack of knowledge. So what was their problem with TB? It is doubtful that anything Mac Life prints has any basis in the technical facts surrounding the issue at hand.
    Apparently, a 'PCIe outside the box' connection tech is due to appear by mid-2014, and will not require controller chips in each cable.
    There have been a number of ways to extend PCI Express over the years, so running PCI Express outside the box isn't exactly new. The trick is to do so reliably.

    As for the controller chips in the cable that gives Apple some flexibility to match the cable to the requirements at hand this plugging in an optical cable is as easy as plugging in a wired cable. The fact remains you need to be able to match the transmission characteristics of the cable to the transmitters, that can be done in cable like Apple does of via something on the logic board.
    Between that and USB 3.1, due in late 2014, Thunderbolt 2 suddenly doesn't seem worth the added developer expense.
    TB 2 is worth every penny if you have a real need for it. The rest of the world should stay with the lower speed busses. Will PCI Express OtB have support for optical cables or other solutions for high reliability long distance transmissions?

    As an aside, don't you just love how "PCI Express Outside the Box" just rolls off your tongue!????????????????????????????????????
    How does this affect the Pro? Having six TB2 ports are one of the machine's big selling points.

    It doesn't affect it one bit. First you have to assume that PCI Express out side the box will take off then you have to assume it will be reliable, neither is a given. Beyond that TB 2 can remain in Apple hardware for as long as required just like other ports have in the past. What the PC industry does has never stopped the Apple industry from doing their own thing.

    In the end the new interface will either be adopted by industry or it won't be adopted. A marginally inferior port might not do to bad but it is hard to tell what it will do in the field. In this respect TB 2/1 is pretty much a proven solution.
  • Reply 1306 of 1320
    drblankdrblank Posts: 3,383member
    Quote:

    Originally Posted by wizard69 View Post





    I don't think you know what you are talking about here. Sadly much of the post I responded to was questions that you asked. If you don't want an opinion or facts why ask the questions.

    Too bad for you. If you are going to ask questions and then reject the answers who here has a problem.

    OK if you say so. Honestly though if I don't make sense then there are an awful lot of people involved in hardware and software development that don't make sense. The goals that the HSAFoundation has come right from their web site. It might not make sense but that is only because there is a lot of confused posting going on out there.

    Well, you are doing a great job at confusing others and in turn, you are confused on your own without anyone else's help.

     

    Seriously, if i need to know more than I currently do, I won't be asking you because you aren't reliable enough for me  to be considered a reliable source of information.

     

    If you want to pass on the existence of HSA to others, that's fine, but until there is some formal announcement that's directly related to Apple, then you might be confusing people.   Don't be part of the confusion.

     

    Remember, Apple most likely already has A8 chips pretty much already set in stone and those get released next year.  A9 chips?  They probably have a good idea what they are going to do with A9 chips already.  Apple has to have actual physically working units way ahead of when they will actually release a product, even though their specs aren't listed anywhere ahead of time.

     

    When a consortium puts together a spec for something, when it is finalized, you won't see it hit the market for at least a year or two.  That cube memory you posted about was being used in Supercomputers.  Apple doesn't make Supercomputers and that memory won't hit the streets in supercomputers until late next year.  If it was meant for desktops, laptops or tablets, I'm sure we'd hear about it well ahead of time.



    So, when there is a credible source that says they are testing this new form of memory or anything for that matter for desktops, laptops and tablets, then we have something to talk about.

  • Reply 1307 of 1320
    frank777frank777 Posts: 5,786member
    Quote:

    Originally Posted by wizard69 View Post





    Haven't seen that, mainly because I can't stomach MacLifes style and lack of knowledge. So what was their problem with TB? It is doubtful that anything Mac Life prints has any basis in the technical facts surrounding the issue at hand.

     

    Can't find a web-link yet, but basically, the feature by Rik Myslewski basically wonders if Apple was shortsighted to put six TB ports on the new Pro, because this 'PCIe outside the box' is coming by mid-2014, and will cost a lot less than TB to implement into peripherals. The article points out that if USB 3.1 and this new PCIe derivative split the market on the PC, TB essentially becomes an expensive, Apple-only tech.

     

    Personally, I can't think of an Apple-adopted interface that only held court for a year, so I have to believe that Apple and Intel thought this through before marching ahead with TB2 this fall. After all, they would have had notice that PCIe OTB was in the works.

     

    But it is sobering to think of the possibility that the new ports on the Pro could be outdated by next July.

  • Reply 1308 of 1320
    drblankdrblank Posts: 3,383member
    Quote:

    Originally Posted by Frank777 View Post

     

     

    Yeah, but PCIe outside the box will offer the speed of Thunderbolt without most of the cost. And it's out mid-2014.

     

    Given that TB2 is just out and it can take a year to develop and market a product, will third-parties support it?

     

    Will TB2 co-exist with PCI OTB?


    Are you sure w/o most of the cost? They made no mention of the cost in that article you mentioned.  or are you referring to the licensing cost?  Licensing cost is the least of the worries.  I think this is mostly for servers anyway.  At least that's what I read in the article upon a quick scan.

     

    One can already get TB PCI slot chassis with and without RAID and some are upgradable to TB 2. Check out NetStor.  

  • Reply 1309 of 1320
    frank777frank777 Posts: 5,786member
    Quote:

    Originally Posted by drblank View Post

     

    Are you sure w/o most of the cost? They made no mention of the cost in that article you mentioned.  or are you referring to the licensing cost?  Licensing cost is the least of the worries.  I think this is mostly for servers anyway.  At least that's what I read in the article upon a quick scan.

     

    One can already get TB PCI slot chassis with and without RAID and some are upgradable to TB 2. Check out NetStor.  


     

    The article mentioned that the new PCI derivative would be a lot cheaper than TB because it would not need chips in the cables themselves (similar, I guess, to USB and FW). That's all I know. Hadn't heard about this new interface before tonight.

  • Reply 1310 of 1320
    drblankdrblank Posts: 3,383member
    Quote:

    Originally Posted by Frank777 View Post

     

     

    The article mentioned that the new PCI derivative would be a lot cheaper than TB because it would not need chips in the cables themselves (similar, I guess, to USB and FW). That's all I know. Hadn't heard about this new interface before tonight.


    There was mention of optical/copper cables as well.  They also mentioned SERVERS, not desktops/laptops.  There are other things like daisy chain able that weren't mentioned.

     

    Seriously, if you are evaluating a product like this and worrying about the price of a cable, then you probably can't afford it.  Cables are the least of your worries.

     

    Trust me, the cheap USB cables out there are usually POS.  They typically aren't well shielded.

     

    If I was going to spend thousands of dollars on an external chassis with other expensive cards, etc. for a production class server or workstation, I'm not buying my cable through Amazon for $10, i can assure you of that.

     

    The electronics inside the TB cables do something that maybe these other PCI things aren't going to do at all, OR they might have other more costly electronics in the actual controller.   Your source wasn't too informed over that, but I assure you, anyone that discusses cable prices and makes that a top priority is probably someone that probably isn't a professional, isn't going to be using it for anything serious and I wouldn't listen to them.

     

    If they are going to have optical and copper wire in the same cable, that might actually be pretty expensive w/o the electronics inside. 

     

    Decent cables cost money.  It's not only making them reliable, but also well shielded so noise from the power side doesn't leak into the data side, etc. because that can cause errors, especially if there are losses with Least Significant Bits.  I even hate using cheap USB cables.  Plus they can crimp too easily and get damaged on the inside.   As they kick up the data speeds, it becomes even more critical to have good quality cables and not the cheap crap they dump on the market.  I'm not a big fan of having to replace cheap cables due to data errors.  How would like to wire up a data center with a ton of cable only to realize that the cable you chose sucks and it causes a lot of network errors, etc. and you have to reachable everything again because you just wanted the cheapest cable you could get because you buy crap on Amazon.

  • Reply 1311 of 1320
    drblankdrblank Posts: 3,383member
    Quote:

    Originally Posted by Frank777 View Post

     

     

    Yeah, but PCIe outside the box will offer the speed of Thunderbolt without most of the cost. And it's out mid-2014.

     

    Given that TB2 is just out and it can take a year to develop and market a product, will third-parties support it?

     

    Will TB2 co-exist with PCI OTB?


    Rik talks about a lot of different things in that article.  Some of it doesn't make sense.  He talks about M-PCIe and three "gears".  And the speeds he's talking about are 1.x Gbps, which isn't that fast.  I don't know what the final end result will be and neither does the author.  One thing about specs for stuff, is that it might actually end up for uses that aren't even related to computers that we use, it might be some other purpose that's more for something completely different.  Like some hand held medical device or something completely different than what we are thinking of.



    For grins, I was on a couple of chip mfg web sites looking at DAC chips and some of them are specifically designed for the auto industry.  They must have had 20 or 30 different DAC chip designs and it was all over the place as to who would use which chip for which purposes.

     

    The main point is this.  Sometimes these specs get thrown around in the pre-release stage and sometimes nothing comes of it, and sometimes it does. Sometimes it eventually winds up in uses that aren't what we originally thought.

     

    Sometimes, I think these authors really don't know how to discuss technology and actually know where, how and all of the other features associated with it.  Yeah, sometimes on paper it looks great, but in reality, it may be a completely different scenario.

     

    You also have cable lengths, can you daisy chain it, etc.  Some of what the author was talking about isn't going to happen until 2015 or even later.

     

    My suggestion, when an author talks about a technology and a big point is over the price of a cable, that's a red flag for you to stop reading and find another article that discusses the technology in a more mature fashion.

  • Reply 1312 of 1320
    nhtnht Posts: 4,429member
    Quote:

    Originally Posted by mdriftmeyer View Post

     

    HSA initiative by AMD and 99% of the industry is taking OpenCL and putting an abstraction layer on top to make the APIs more `user friendly' and IDE ready.

     


     

    This is key.  Well this and using OpenCL behind the scenes in the exiting APIs without much effort on the part of developers.  High parallelism is difficult as is coding for OpenCL in comparison traditional coding.

  • Reply 1313 of 1320
    wizard69wizard69 Posts: 12,755member
    nht wrote: »
    This is key.  Well this and using OpenCL behind the scenes in the exiting APIs without much effort on the part of developers.  High parallelism is difficult as is coding for OpenCL in comparison traditional coding.

    There are some misunderstandings about AMDs long term goals for the HSA initiative. Long term OpenCL will be one way of many to leverage GPU compute. If you look at the organizations slide you will see that they have plans to eliminate OpenCL for the Java implementation even if Java will be layered upon OpenCL for a couple of years. The ultimate goal is to have Java communicate with HSAIL directly taking OpenCL out of the loop, that might be 2016 though. Other languages may end up bypassing OpenCL also, HSAIL has the ability to support C++ for example.

    This doesn't mean that OpenCL goes away, it just becomes one way to exercise the HSA facilities coming in this new hardware. HSAIL is in fact a bit like LLVM and the way it is used in the system the HSA Foundation is putting together. I don't want to say they copied Apple but the concept is extremely similar.
  • Reply 1314 of 1320

    Has Relic posted lately? It's been 2 or 3 weeks since I remember seeing her last one.

  • Reply 1315 of 1320
    Has Relic posted lately? It's been 2 or 3 weeks since I remember seeing her last one.

    Still around! Seen last Friday, but posts little.
  • Reply 1316 of 1320
    wizard69wizard69 Posts: 12,755member
    philboogie wrote: »
    Still around! Seen last Friday, but posts little.

    I'm hoping she pulls through!
  • Reply 1317 of 1320
    hmmhmm Posts: 3,405member
    Quote:

    Originally Posted by wizard69 View Post





    I'm hoping she pulls through!



    no posts in the last few days :(

  • Reply 1318 of 1320

    Last post was about game controllers:

     

    post #6

     

     

    Hope to see her post again soon...

  • Reply 1319 of 1320
    wizard69wizard69 Posts: 12,755member
    hmm wrote: »

    no posts in the last few days :(

    Last post was about game controllers:

    post #6


    Hope to see her post again soon...

    This is a terribly sad situation. Way too young. Hopefully we do see a new post from her soon
  • Reply 1320 of 1320
    wizard69 wrote: »
    This is a terribly sad situation. Way too young. Hopefully we do see a new post from her soon

    Fully agree. And her sense of humour! So much hope she can pull through. Will hit me hard if I see her obituary here.
Sign In or Register to comment.