I suspect discrete gpus will become more expensive, and NVidia will need some way to absorb fabrication costs.
Right now higher prices for discrete GPUs seem to be the only possibility.
Right now even the gaming Fermi cards are still so far ahead of integrated graphics in areas that don't involve gaming, such as computation or OpenGL heavy apps. Even if every new product involved an integrated gpu starting today, it would take some time for them to filter through.
True today but maybe not tomorrow. AMD is doing wonders with their APUs and GPU performance. It will take time but maybe by 2014 we will have very good integrated GPUs even for modest gaming.
The slight upside to this is that it would drive improvements in that area. Intel clearly doesn't see it going 100% integrated just yet. Their E5s/EP cpu packages do not ship with integrated gpus. EN has been shipping with integrated at least since Sandy. NVidia hasalready cut out some of their development partners in favor of producing cards directly. Tom's hardware had some speculation that they'll try to implement ARM in their Tesla solutions. They obviously have some kind of plan, but I'm not knowledgeable enough in terms of hardware development to offer any real insight.
NVidia is probably after the extra value of. Complete cards. It is the only way to to help defray development costs.
I don't entirely agree there. Fermi blew everything else away in double precision math, and NVidia was the one that really pioneered GPGPU.
I'm not sure where you get this idea, for the most part NVidia sucks at double precision. AMD has a wide array of chips supporting double precision.
OpenCL came along after, and in some applications NVidia is still preferred. It's just much less so on OSX.
OpenCL is getting very wide industry acceptance. This isn't just an Mac OS/X initiative.
They're actually quite strong in that area. They even have an entire gpu line with a huge amount of on board ram dedicated to such functions with Tesla computing. Some of AMD's driver problems do not exist in OSX, which is probably a reason for the differing balance of power. I suspect the market for performance machines will hold out for a number of years.
Yes I agree power machines will basically be around forever. I wasn't implying that the machines wouldn't be around but rather that it is hard to see the likes of Dell or HP surviving on just performance workstations. Frankly all of the workstation manufactures died off a long time ago.
If the primary focus in consumer technology is on mobile devices with minimized power consumption, it will leave a gap in the market for something that offers more performance beyond just a thin client hooked up to a server farm. HP seems quite happy with themselves when it comes to workstations, which isn't the segment they considered spinning off. They make too much on them to even consider that. I wonder how the Z1 has worked out for them.
Well maybe we will learn soon. Funny thing is HPs quarterly aren't news like Apples.
HP seems quite happy with themselves when it comes to workstations, which isn't the segment they considered spinning off. They make too much on them to even consider that. I wonder how the Z1 has worked out for them.
They might contribute more profit than any of their lineup but the numbers are still low:
Q4 2011 not even 1 million shipped worldwide with HP still at 41% and Dell at 33%. Apple has 1/3 the shipments of these so if we assume it's the same here, they'd have 10% = 100,000 units per quarter. That matches with the 5% estimate of their lineup, which is about 75,000 units.
If every workstation HP sold was $5000, they'd make $2000 per unit (gross) x 400,000 = $800m. Their net profit is usually around $1.5b (when they aren't writing off massive losses) and their net margins are about 1/3 of the gross:
So net profit from their workstation line (assuming the average workstation is $5000, which is probably optimistic) is about $250m / quarter out of $1.5b / quarter. It's a good profit margin but it is countered by the extremely low volume. HP ships about 15m machines per quarter so 400,000 is less than 3% of their lineup and I expect it will be around the same for Apple.
If Apple even made the same profit HP did with 1/3 of the sales, they'd make $250m out of $8b = 3% of their profit. Every portion helps out but it's a tiny fraction.
If the primary focus in consumer technology is on mobile devices with minimized power consumption, it will leave a gap in the market for something that offers more performance beyond just a thin client hooked up to a server farm.
The 27" iMac still has a 310W PSU - you seem to be suggesting that lower-end machines will all end up at a very low power draw but they scale up in power the higher up they go and this gap in the market will stay very, very small.
If Apple even made the same profit HP did with 1/3 of the sales, they'd make $250m out of $8b = 3% of their profit. Every portion helps out but it's a tiny fraction.
So by this reasoning, Chevrolet should kill the Corvette and Dodge should kill the Viper. Neither car amounts to much in profits, so what good are they?
Of course, if Chevy killed the Corvette, it wouldn't affect the rest of their lineup's functionality. If Apple kills the Mac Pro, eventually there will be less interest in Mac OS X among content creation developers, and so the iMac will be able to do less.
If Apple kills the Mac Pro, eventually there will be less interest in Mac OS X among content creation developers, and so the iMac will be able to do less.
That's a pretty big assumption. The difference with cars is that even consumers admire the iconic models as being the better models. With computers, the iMac, MBP, iPad, iPhone by far get the most admiration with the MP all but forgotten about.
I would bet Adobe sells more copies of their CS Suite to iMac owners than MP owners. More specialised software like Houdini, Nuke, Maya, AutoCAD etc could sell more to the workstation crowd but they are more recently targeting the higher volume computer models and there's a huge student market.
Another problem lies in the belief that without a MP, the entire MP audience would abandon the platform. OS X offers too much value to give up on it for the sake of raw horsepower, which can be gained by other means.
I know you want to keep tying content creators and high-end software to the MP but that exclusivty is fictitious.
None of that means they should discontinue anything, just that they could. We know that they aren't for now.
So by this reasoning, Chevrolet should kill the Corvette and Dodge should kill the Viper. Neither car amounts to much in profits, so what good are they?
I don't think that was his point. The key with these cars especially the Corvette is that they have sales strong enough to cover developmental costs and production costs. With the Mac Pro it isn't even clear that they cover those costs.
Of course, if Chevy killed the Corvette, it wouldn't affect the rest of their lineup's functionality. If Apple kills the Mac Pro, eventually there will be less interest in Mac OS X among content creation developers, and so the iMac will be able to do less.
If Apple kills the Mac Pro many professionals would give up on Mac OS all together. This would be the case even if they have never used a Mac Pro. The Mac Pro demonstrates to many that Apple supports professional users.
In any event to bring this thread back around to the Mini, it is Apples wishy washy approach to discrete GPUs that is really under minding their credibility with not only professional but those users needing a bit more than baseline performance out of their computers. In my case I'm stuck between a Mini that isn't even close to powerful enough for a long term investment and a Mac Pro which is way to expensive for what you get.
So if you are wondering what I'm wishing for, it is a desktop box, Mini or Pro with respectable mid range performance. That means a decent discrete GPU these days with a suitable amount of VRAM. The 2011 Mini didn't provide that and the 2012 Mini abandoned discrete GPUs altogether. Frankly I'm not sure why Apple has such a hard time grasping the need for a decently priced midrange machine. Such a machine should be able to be comfortably priced between the Mini and Mac Pro and could be built on the chassis of either as far as I'm concerned. In the case of the Mini add an inch of height to the case if needed to support a reasonable GPU if it is needed.
When I purchased my MBP in 2008 I had been waiting over three years to see Apple revamp thier lineup to deliver a decent desktop machine. Eventually I said screw it and purchased the MBP. Now it is five years latter and Apple has done nothing to improve the desktop lineup. That is depressing. I still beleive that nobody at Apple gives a damn about the desktop line. I mean really it has been more than 8 years of terrible values in the desktop line up. The Mini went one step forward and two back this year.
Base - Dual core/4 GB RAM/Intel HD 4600/500 GB HDD
I don't like this configuration due to far to little RAM for a machine shipping in 2013. Also quad cores will soon be baseline configurations.
Upgraded - Higher dual core or quad core/4 GB RAM/Intel HD 4600/1 TB HDD
Did you give upon your desire for a discrete GPU? Honestly this should be the defining feature of this model.
I have the same issue here with base RAM, 4GB is just not enough for 2013.
Server - Quad core/4 GB RAM/Intel HD 4600/2x1 TB HDD
The so called server is just marketing BS. Given that you still are configuring with far to little RAM, especially for a server. I'd actually like to see this machine designed for compute intensive users. That is put the fastest quad or six core processor in the machine that you can. The reality is the "server" is a good machine for developers but XCode is hard on hardware and needs plenty of fast threads. So lots of fast threads need to be the Mini "servers" forte.
I'm hoping at least the upgraded has discrete graphics though no cards are out so I can't predict the card to be used.
iMac I need to see the cards nVidia releases before I judge there too.
Personally I'm hoping for an AMD GPU. Sadly I think this is the end of discrete GPUs in the Mini but I suppose we can keep asking. Like I've said before I really don't know what in the hell Apple is doing with its desktop line up. I can see a complete refresh of the Mini timed with the debut of the new Mac Pro.
I don't know what 2.5" drives are in the Mini/MBP, and I don't have the link in front of me, but Seagate is stopping production of pure 2.5" HDD and replacing them with new versions of their Momentous XT type of drives. In other words, the drives will all have flash mixed with spinning platters for traditional HDD.
I don't know what 2.5" drives are in the Mini/MBP, and I don't have the link in front of me, but Seagate is stopping production of pure 2.5" HDD and replacing them with new versions of their Momentous XT type of drives. In other words, the drives will all have flash mixed with spinning platters for traditional HDD.
I'm sure it's only Seagate and Western Digital left in the HDD market and WD is doing a hybrid drive too.
So to compete on price, Seagate will probably do the same. In a Fusion setup, that'll be a double cache going on. Plus how many people will opt for Fusion if they get a cache mechanism anyway? Fusion will work better with more storage but the 24GB would fit the OS and some apps and you'll get the write caching. It's good for consumers anyway.
If Apple even made the same profit HP did with 1/3 of the sales, they'd make $250m out of $8b = 3% of their profit. Every portion helps out but it's a tiny fraction.
The 27" iMac still has a 310W PSU - you seem to be suggesting that lower-end machines will all end up at a very low power draw but they scale up in power the higher up they go and this gap in the market will stay very, very small.
I didn't suggest this was a market where Apple would be concerned. I merely stated that the margins are high enough that some companies will continue serving it going forward. Beyond that I was saying if Apple wanted to maintain it, they'd likely try to gain customers from these other brands. I never suggested it would be either a big market for them or a prime focus within their marketing. I just said that without leeching customers from other vendors, their portion of that market is likely tapped out on growth at this point. Both of these things are independent of their internal decisions.
Quote:
Originally Posted by wizard69
Right now higher prices for discrete GPUs seem to be the only possibility.
True today but maybe not tomorrow. AMD is doing wonders with their APUs and GPU performance. It will take time but maybe by 2014 we will have very good integrated GPUs even for modest gaming.
NVidia is probably after the extra value of. Complete cards. It is the only way to to help defray development costs.
Quote:
I'm not sure where you get this idea, for the most part NVidia sucks at double precision. AMD has a wide array of chips supporting double precision.
OpenCL is getting very wide industry acceptance. This isn't just an Mac OS/X initiative.
Kepler numbers were kind of weak. NVidia can handle both CUDA and OpenCL applications. What I mentioned was simply that CUDA gained some traction earlier on, so the code was out there. It may very well be a dead end within 3 years, but today it's important to look at what applications you are using prior to making the decision. I correct others frequently on false assumptions of what the gpu will and will not accelerate in specific applications, as many don't like to read, no matter how easy the developer makes it.
Quote:
Yes I agree power machines will basically be around forever. I wasn't implying that the machines wouldn't be around but rather that it is hard to see the likes of Dell or HP surviving on just performance workstations. Frankly all of the workstation manufactures died off a long time ago.
Yes.... no more $30k Quantel or SGI workstations. No more turnkey systems. I get that. Desktop hardware and software development outpaced them. Some of the really cool advancements are being driven by phones today. You have a huge amount of profit in that area, and it's driving R&D. I'm actually somewhat conservative on phone upgrades. I upgrade when there's a big jump in capability relative to how I use it, or the camera shows major improvements.
Quote:
Well maybe we will learn soon. Funny thing is HPs quarterly aren't news like Apples.
I never suggested that. Some of these other oems are fairly dependent on wherever they can bring in high margins, regardless of volume. Their consumer lines that carry the bulk of the volume are often nearly devoid of profits after accounting for R&D and support. Apple doesn't really have that problem.
So to compete on price, Seagate will probably do the same. In a Fusion setup, that'll be a double cache going on. Plus how many people will opt for Fusion if they get a cache mechanism anyway? Fusion will work better with more storage but the 24GB would fit the OS and some apps and you'll get the write caching. It's good for consumers anyway.
I've had much better luck with Western Digital. Seagate shows up in oem builds quite a lot due to price. They remain quite healthy in spite of their last man standing appearance. They retain a presence in enterprise grade solutions, and it's likely that they have a strategy in mind going forward.
That's weird, I could swear I made a post bolding some points that wizard69 made on my predictions for the 2013 Mini.
Anyway, I will keep what I said before brief since I'm too lazy to type it again.
- Perhaps kill the server model unless it is a big seller. - Make 8 GB of RAM the standard for the $799 model (I'd like to see it as well for $599 model but Apple would never do it) - Discrete graphics in the $799 model with at least 512 MB or even 1 GB although Apple would probably settle on 768 MB before 1 GB - Quad cores across the board would be nice but they'll always go dual for the low end.
I'll really stretch out here and say knock $100 off across the board but especially the base mini.
And commercials... I know SOMEBODY could make a clever commercial for the Super Bowl for the Mac mini and it would drive sales tremendously.
- Perhaps kill the server model unless it is a big seller.
I'm not sure why you're concerned with that. Do you think it's somehow holding the mini back? The server model is just clever marketing. It's unlikely that it requires much in the way of validation. People were using the mini with OSX server, so Apple thought they could prop up sales on it by slapping that label on one model. It doesn't constrict the lineup in any way.
From the user perspective a disk with a flash cache may give the same results as a Fusion drive but technically we are talking about two different things here. I'm not sure if Apple will be able to successfully market Fusion drive against magnetic drives with flash caches but right now if I had a choice I'd go with Fusion drive. If they expanded the capability of the Fusion software to support larger that 128GB SSDs that would put a lock on Fusion drive tech.
So to compete on price, Seagate will probably do the same. In a Fusion setup, that'll be a double cache going on. Plus how many people will opt for Fusion if they get a cache mechanism anyway? Fusion will work better with more storage but the 24GB would fit the OS and some apps and you'll get the write caching. It's good for consumers anyway.
I'm not sure why you're concerned with that. Do you think it's somehow holding the mini back? The server model is just clever marketing. It's unlikely that it requires much in the way of validation. People were using the mini with OSX server, so Apple thought they could prop up sales on it by slapping that label on one model. It doesn't constrict the lineup in any way.
Brainstorming or perhaps lack thereof. Put all of the focus on the discrete model and make the high end model and make the base model $100 less. Put the server options (for HDDs and SSDs) as BTOs in the high end model.
Brainstorming or perhaps lack thereof. Put all of the focus on the discrete model and make the high end model and make the base model $100 less. Put the server options (for HDDs and SSDs) as BTOs in the high end model.
I think you misinterpret the focus of the machine. If it was a one of their bigger sellers, it would have a display option in better alignment with its price.
You're too hung up on that. The parts used aren't drastically different in any of them. They label one a server to drive extra sales. The idea originated with their customer base. They put it in there to drive extra sales. As to a discrete gpu, I don't see it ever coming back. They made a mediocre effort one round. They dropped it the next. This means they expect further improvements will ensure performance that meets the criteria of good enough. Apple has gone this route before. They'll push machines with higher price points as their performance options. 8GB of ram should have already been the norm, but it's cheap enough to upgrade after market. You can do that for under $50. 16 is only $80 or so. They should provide cto drive options throughout the range rather than forcing the purchase of higher end models to access such things. That is a typical way to prop up margins, but I hope it backfires on them with people doing their own upgrades. When you look at the total cost to get a Fusion drive, some people may opt for an after market drive. I know I've seen Samsung 830s 512GB models fall below the $500 mark. That doesn't help if you want terabytes of storage, but you may want to opt for Applecare if you go that route. At the very least you'd want to find out what the replacement cost is for a Fusion drive, as it's no more immortal than any other HDD or SSD.
You're too hung up on that. The parts used aren't drastically different in any of them.
This is why I find Apples marketing so frustrating, the server model isn't drastically different from the base model, most likely varying only in the processor implemented. It is a totally artificial distinction. Even more frustrating is maintains options based on machine designation.
They label one a server to drive extra sales. The idea originated with their customer base. They put it in there to drive extra sales. As to a discrete gpu, I don't see it ever coming back. They made a mediocre effort one round. They dropped it the next.
Mediocre is being kind. The 2011 model was an ample demonstration that Apple doesn't even grasp what their customers are asking for. The whole idea behind asking for a GPU in the Mini is to get good compute and 3D performance and support for modern software. The 2011 GPU model failed basic customer requirements right out of the box.
This means they expect further improvements will ensure performance that meets the criteria of good enough. Apple has gone this route before. They'll push machines with higher price points as their performance options. 8GB of ram should have already been the norm, but it's cheap enough to upgrade after market.
That excuse of being cheap enough doesn't support Apple idiocy here. In fact it highlights just how stingy Apple is with RAM. Even if the base model stays at 4GB hold the rest of the line at 4GB is just stupid. The sad reality is that Apples Minis become poorer values as they get more expensive.
You can do that for under $50. 16 is only $80 or so. They should provide cto drive options throughout the range rather than forcing the purchase of higher end models to access such things. That is a typical way to prop up margins, but I hope it backfires on them with people doing their own upgrades.
Frankly I think people have had it with Apples pricing structure, at least when it comes to the Mini. That is likely a factor in lackluster Mini sales.
When you look at the total cost to get a Fusion drive, some people may opt for an after market drive. I know I've seen Samsung 830s 512GB models fall below the $500 mark. That doesn't help if you want terabytes of storage, but you may want to opt for Applecare if you go that route. At the very least you'd want to find out what the replacement cost is for a Fusion drive, as it's no more immortal than any other HDD or SSD.
I suppose the best option for drive setup varies with users. However I wold likely go with an SSD for system & apps and put data on a magnetic drive if I get a Mini this year. As you note SSDs are dropping in price fairly quickly thus such an arraignment would work out well for many users. In this regard I do wish that the Mini was shipped configured with cable and parts to make DIY installs easy. Last I knew anyways the base Mini didn't come with the hardware to allow quick and easy installation of a second drive.
As to the replacement cost of a "Fusion Drive", remember it is just two drives tied together with some software volume management tricks. You can already DIY a Fusion solution.
Comments
They might contribute more profit than any of their lineup but the numbers are still low:
http://www.xbitlabs.com/news/other/display/20120322235407_Nearly_Million_of_Workstations_Shipped_in_Fourth_Quarter_Market_Researchers.html
Q4 2011 not even 1 million shipped worldwide with HP still at 41% and Dell at 33%. Apple has 1/3 the shipments of these so if we assume it's the same here, they'd have 10% = 100,000 units per quarter. That matches with the 5% estimate of their lineup, which is about 75,000 units.
Here's an HP workstation parts:
E5-2680 = $1723
FirePro V4900 = $155
4GB RAM = $40
500GB HDD = $60
Keyboard/mouse = $40
Intel C602 motherboard = $600
1125W PSU = $150
enclosure = $200?
wifi/bluetooth = $100
Parts = $3068
HP Z820 Price = $5053
Margin = 40%
If every workstation HP sold was $5000, they'd make $2000 per unit (gross) x 400,000 = $800m. Their net profit is usually around $1.5b (when they aren't writing off massive losses) and their net margins are about 1/3 of the gross:
http://www.stock-analysis-on.net/NYSE/Company/Hewlett-Packard-Co/Ratios/Profitability
So net profit from their workstation line (assuming the average workstation is $5000, which is probably optimistic) is about $250m / quarter out of $1.5b / quarter. It's a good profit margin but it is countered by the extremely low volume. HP ships about 15m machines per quarter so 400,000 is less than 3% of their lineup and I expect it will be around the same for Apple.
If Apple even made the same profit HP did with 1/3 of the sales, they'd make $250m out of $8b = 3% of their profit. Every portion helps out but it's a tiny fraction.
The 27" iMac still has a 310W PSU - you seem to be suggesting that lower-end machines will all end up at a very low power draw but they scale up in power the higher up they go and this gap in the market will stay very, very small.
Quote:
Originally Posted by Marvin
If Apple even made the same profit HP did with 1/3 of the sales, they'd make $250m out of $8b = 3% of their profit. Every portion helps out but it's a tiny fraction.
So by this reasoning, Chevrolet should kill the Corvette and Dodge should kill the Viper. Neither car amounts to much in profits, so what good are they?
Of course, if Chevy killed the Corvette, it wouldn't affect the rest of their lineup's functionality. If Apple kills the Mac Pro, eventually there will be less interest in Mac OS X among content creation developers, and so the iMac will be able to do less.
That's a pretty big assumption. The difference with cars is that even consumers admire the iconic models as being the better models. With computers, the iMac, MBP, iPad, iPhone by far get the most admiration with the MP all but forgotten about.
I would bet Adobe sells more copies of their CS Suite to iMac owners than MP owners. More specialised software like Houdini, Nuke, Maya, AutoCAD etc could sell more to the workstation crowd but they are more recently targeting the higher volume computer models and there's a huge student market.
Another problem lies in the belief that without a MP, the entire MP audience would abandon the platform. OS X offers too much value to give up on it for the sake of raw horsepower, which can be gained by other means.
I know you want to keep tying content creators and high-end software to the MP but that exclusivty is fictitious.
None of that means they should discontinue anything, just that they could. We know that they aren't for now.
Base - Dual core/4 GB RAM/Intel HD 4600/500 GB HDD
Upgraded - Higher dual core or quad core/4 GB RAM/Intel HD 4600/1 TB HDD
Server - Quad core/4 GB RAM/Intel HD 4600/2x1 TB HDD
I'm hoping at least the upgraded has discrete graphics though no cards are out so I can't predict the card to be used.
iMac I need to see the cards nVidia releases before I judge there too.
If Apple kills the Mac Pro many professionals would give up on Mac OS all together. This would be the case even if they have never used a Mac Pro. The Mac Pro demonstrates to many that Apple supports professional users.
In any event to bring this thread back around to the Mini, it is Apples wishy washy approach to discrete GPUs that is really under minding their credibility with not only professional but those users needing a bit more than baseline performance out of their computers. In my case I'm stuck between a Mini that isn't even close to powerful enough for a long term investment and a Mac Pro which is way to expensive for what you get.
So if you are wondering what I'm wishing for, it is a desktop box, Mini or Pro with respectable mid range performance. That means a decent discrete GPU these days with a suitable amount of VRAM. The 2011 Mini didn't provide that and the 2012 Mini abandoned discrete GPUs altogether. Frankly I'm not sure why Apple has such a hard time grasping the need for a decently priced midrange machine. Such a machine should be able to be comfortably priced between the Mini and Mac Pro and could be built on the chassis of either as far as I'm concerned. In the case of the Mini add an inch of height to the case if needed to support a reasonable GPU if it is needed.
When I purchased my MBP in 2008 I had been waiting over three years to see Apple revamp thier lineup to deliver a decent desktop machine. Eventually I said screw it and purchased the MBP. Now it is five years latter and Apple has done nothing to improve the desktop lineup. That is depressing. I still beleive that nobody at Apple gives a damn about the desktop line. I mean really it has been more than 8 years of terrible values in the desktop line up. The Mini went one step forward and two back this year.
I don't like this configuration due to far to little RAM for a machine shipping in 2013. Also quad cores will soon be baseline configurations. Did you give upon your desire for a discrete GPU? Honestly this should be the defining feature of this model.
I have the same issue here with base RAM, 4GB is just not enough for 2013. The so called server is just marketing BS. Given that you still are configuring with far to little RAM, especially for a server. I'd actually like to see this machine designed for compute intensive users. That is put the fastest quad or six core processor in the machine that you can. The reality is the "server" is a good machine for developers but XCode is hard on hardware and needs plenty of fast threads. So lots of fast threads need to be the Mini "servers" forte. Personally I'm hoping for an AMD GPU. Sadly I think this is the end of discrete GPUs in the Mini but I suppose we can keep asking. Like I've said before I really don't know what in the hell Apple is doing with its desktop line up. I can see a complete refresh of the Mini timed with the debut of the new Mac Pro.
I don't know what 2.5" drives are in the Mini/MBP, and I don't have the link in front of me, but Seagate is stopping production of pure 2.5" HDD and replacing them with new versions of their Momentous XT type of drives. In other words, the drives will all have flash mixed with spinning platters for traditional HDD.
I'm sure it's only Seagate and Western Digital left in the HDD market and WD is doing a hybrid drive too.
WD is using 16-24GB NAND in their new ones:
http://reviews.cnet.com/hard-drives/wd-black-sshd-1tb/4505-3186_7-35567266.html
So to compete on price, Seagate will probably do the same. In a Fusion setup, that'll be a double cache going on. Plus how many people will opt for Fusion if they get a cache mechanism anyway? Fusion will work better with more storage but the 24GB would fit the OS and some apps and you'll get the write caching. It's good for consumers anyway.
Quote:
Originally Posted by Marvin
They might contribute more profit than any of their lineup but the numbers are still low:
http://www.xbitlabs.com/news/other/display/20120322235407_Nearly_Million_of_Workstations_Shipped_in_Fourth_Quarter_Market_Researchers.html
You find some of the best links.
Quote:
If Apple even made the same profit HP did with 1/3 of the sales, they'd make $250m out of $8b = 3% of their profit. Every portion helps out but it's a tiny fraction.
The 27" iMac still has a 310W PSU - you seem to be suggesting that lower-end machines will all end up at a very low power draw but they scale up in power the higher up they go and this gap in the market will stay very, very small.
I didn't suggest this was a market where Apple would be concerned. I merely stated that the margins are high enough that some companies will continue serving it going forward. Beyond that I was saying if Apple wanted to maintain it, they'd likely try to gain customers from these other brands. I never suggested it would be either a big market for them or a prime focus within their marketing. I just said that without leeching customers from other vendors, their portion of that market is likely tapped out on growth at this point. Both of these things are independent of their internal decisions.
Quote:
Originally Posted by wizard69
Right now higher prices for discrete GPUs seem to be the only possibility.
True today but maybe not tomorrow. AMD is doing wonders with their APUs and GPU performance. It will take time but maybe by 2014 we will have very good integrated GPUs even for modest gaming.
NVidia is probably after the extra value of. Complete cards. It is the only way to to help defray development costs.
Quote:
I'm not sure where you get this idea, for the most part NVidia sucks at double precision. AMD has a wide array of chips supporting double precision.
OpenCL is getting very wide industry acceptance. This isn't just an Mac OS/X initiative.
Kepler numbers were kind of weak. NVidia can handle both CUDA and OpenCL applications. What I mentioned was simply that CUDA gained some traction earlier on, so the code was out there. It may very well be a dead end within 3 years, but today it's important to look at what applications you are using prior to making the decision. I correct others frequently on false assumptions of what the gpu will and will not accelerate in specific applications, as many don't like to read, no matter how easy the developer makes it.
Quote:
Yes I agree power machines will basically be around forever. I wasn't implying that the machines wouldn't be around but rather that it is hard to see the likes of Dell or HP surviving on just performance workstations. Frankly all of the workstation manufactures died off a long time ago.
Yes.... no more $30k Quantel or SGI workstations. No more turnkey systems. I get that. Desktop hardware and software development outpaced them. Some of the really cool advancements are being driven by phones today. You have a huge amount of profit in that area, and it's driving R&D. I'm actually somewhat conservative on phone upgrades. I upgrade when there's a big jump in capability relative to how I use it, or the camera shows major improvements.
Quote:
Well maybe we will learn soon. Funny thing is HPs quarterly aren't news like Apples.
I never suggested that. Some of these other oems are fairly dependent on wherever they can bring in high margins, regardless of volume. Their consumer lines that carry the bulk of the volume are often nearly devoid of profits after accounting for R&D and support. Apple doesn't really have that problem.
Quote:
Originally Posted by Marvin
I'm sure it's only Seagate and Western Digital left in the HDD market and WD is doing a hybrid drive too.
WD is using 16-24GB NAND in their new ones:
http://reviews.cnet.com/hard-drives/wd-black-sshd-1tb/4505-3186_7-35567266.html
So to compete on price, Seagate will probably do the same. In a Fusion setup, that'll be a double cache going on. Plus how many people will opt for Fusion if they get a cache mechanism anyway? Fusion will work better with more storage but the 24GB would fit the OS and some apps and you'll get the write caching. It's good for consumers anyway.
I've had much better luck with Western Digital. Seagate shows up in oem builds quite a lot due to price. They remain quite healthy in spite of their last man standing appearance. They retain a presence in enterprise grade solutions, and it's likely that they have a strategy in mind going forward.
Anyway, I will keep what I said before brief since I'm too lazy to type it again.
- Perhaps kill the server model unless it is a big seller.
- Make 8 GB of RAM the standard for the $799 model (I'd like to see it as well for $599 model but Apple would never do it)
- Discrete graphics in the $799 model with at least 512 MB or even 1 GB although Apple would probably settle on 768 MB before 1 GB
- Quad cores across the board would be nice but they'll always go dual for the low end.
I'll really stretch out here and say knock $100 off across the board but especially the base mini.
And commercials... I know SOMEBODY could make a clever commercial for the Super Bowl for the Mac mini and it would drive sales tremendously.
Quote:
Originally Posted by Winter
- Perhaps kill the server model unless it is a big seller.
I'm not sure why you're concerned with that. Do you think it's somehow holding the mini back? The server model is just clever marketing. It's unlikely that it requires much in the way of validation. People were using the mini with OSX server, so Apple thought they could prop up sales on it by slapping that label on one model. It doesn't constrict the lineup in any way.
Quote:
Originally Posted by Marvin
...It's good for consumers anyway.
Apple's corporate motto.
Brainstorming or perhaps lack thereof. Put all of the focus on the discrete model and make the high end model and make the base model $100 less. Put the server options (for HDDs and SSDs) as BTOs in the high end model.
Quote:
Originally Posted by Winter
Brainstorming or perhaps lack thereof. Put all of the focus on the discrete model and make the high end model and make the base model $100 less. Put the server options (for HDDs and SSDs) as BTOs in the high end model.
I think you misinterpret the focus of the machine. If it was a one of their bigger sellers, it would have a display option in better alignment with its price.
Quote:
Originally Posted by Winter
Three models? Two models? How do you handle it?
You're too hung up on that. The parts used aren't drastically different in any of them. They label one a server to drive extra sales. The idea originated with their customer base. They put it in there to drive extra sales. As to a discrete gpu, I don't see it ever coming back. They made a mediocre effort one round. They dropped it the next. This means they expect further improvements will ensure performance that meets the criteria of good enough. Apple has gone this route before. They'll push machines with higher price points as their performance options. 8GB of ram should have already been the norm, but it's cheap enough to upgrade after market. You can do that for under $50. 16 is only $80 or so. They should provide cto drive options throughout the range rather than forcing the purchase of higher end models to access such things. That is a typical way to prop up margins, but I hope it backfires on them with people doing their own upgrades. When you look at the total cost to get a Fusion drive, some people may opt for an after market drive. I know I've seen Samsung 830s 512GB models fall below the $500 mark. That doesn't help if you want terabytes of storage, but you may want to opt for Applecare if you go that route. At the very least you'd want to find out what the replacement cost is for a Fusion drive, as it's no more immortal than any other HDD or SSD.
As to the replacement cost of a "Fusion Drive", remember it is just two drives tied together with some software volume management tricks. You can already DIY a Fusion solution.