Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

hishnash

Member
  • Content Count

    623
  • Joined

  • Last visited

Reputation Activity

  1. Like
    hishnash reacted to TheReal1980 in MateBook X Pro vs M1 MacBook Air - Pretty insane...   
    Again...
    Photoshop Super Resolution Test
    Apple: 13 seconds.
    Huawei: 10+ minutes.
     
    48 TIMES FASTER...
     
    I wonder if we could use this tech in games in the future, would that be possible? Either way it is very interesting.
  2. Like
    hishnash reacted to TheReal1980 in Surface Laptop (Intel and AMD) vs M1 MacBook Air. Oh my...   
    I would like Linus and the gang to inspect this much further with real life situations.
    How many times can you render a specific movie clip for example by only using the battery? Does each rendered movie clip take longer with time? How long did it take to render it 20 times et.c.
  3. Like
    hishnash reacted to Zodiark1593 in Surface Laptop (Intel and AMD) vs M1 MacBook Air. Oh my...   
    Calling the M1, Apple Silcon is an entirely fair assessment. Apple doesn’t license ARM cores, but rather, the ISA. The CPU cores are 100% Apple’s design that is compatible with the ARMv8 ISA, and in fact, is known that Apple has added at least one extension set (Apple AMX) that doesn’t exist in ARM’s ISA. 
  4. Like
    hishnash reacted to Spindel in Surface Laptop (Intel and AMD) vs M1 MacBook Air. Oh my...   
    I've naged about this in a lot of threads about the M1. 
     
    As an owner of an M1 Mini one thing no benchmark can't communicate and most (if not all reviews) do not communicate is the over all responsiveness of the system when using it. I understand its omitted because it is a thing that is hard to quantify, but no computer I have ever used before has the same feeling of responsiveness when actually using the computer. 
  5. Like
    hishnash reacted to mon1ka in It's Finally Out! iOS/iPadOS 14.5 has been released with the ability to block app (Facebook) tracking.   
    good. glad apple is actually taking action.
     
    This is BS IMO. if facebook really cared about small buissness, why are they only now defending something 'that helps them?' seems like marketing buzz IMO
  6. Like
    hishnash reacted to JLO64 in It's Finally Out! iOS/iPadOS 14.5 has been released with the ability to block app (Facebook) tracking.   
    Summary
     iOS/iPadOS 14.5 has just been released this Monday with its biggest feature called "App Tracking Transparency" which will allow users to block apps from tracking them across other apps and websites which will seriously affect the ability for ad companies like Facebook and Google to serve personalized ads to users. Users will be prompted whether or not they want to allow a specific app to track them when it attempts to do so.  In this same prompt developers are given a paragraph to explain why users should allow themselves to be tracked. Additional settings to block app tracking are also available within the settings app. Apple has reportedly held back this update in order to allow developers to prepare for it and because of the backlash they faced from several companies, most notably Facebook.
     
    Last year Facebook took out a full page ad in the Wall Street Journal proclaiming that this feature would hurt small businesses more than anybody else.
    Apple replied to this ad with this statement:
     
    Other companies have publicly reacted to this feature (just not as loud as Facebook did) with Unity and Snapchat warning their respective investors that they will be impacted.
     
    Finally, Apple has stated that they will punish developers who track users without permission.
     
    My thoughts
    It'll be interesting to see how Facebook and Google will react to these changes in the long run as this will impact core parts of their advertising business. While iPhone users don't have a market share as big as Android users, they are more likely on average to spend more than their counterparts. There will be a dent to both of these companies and they know it. What's most surprising to me is how little Google has been fighting this. My guess is that they won't be as affected by this as Facebook is since plenty of iPhone users use their services and apps (Google is the default search engine for Apple devices) and Apple can't block Google from tracking them there.
     
    In any case once I finish posting this I'll be updating my iPad to it. I will be interested in seeing how the user experience is like for blocking app tracking.
     
    Oh they also added the ability to unlock a Face ID equipped phone with a mask on if you have an Apple Watch, support for Apple's new Airtags, updated the podcasts app (I hate Spotify's podcast UI with a burning passion so yay), support for Xbox Series and PS5 controllers, and the ability to set Spotify as the default music player when using Siri.
     
    Sources
    https://www.inc.com/jason-aten/facebook-took-out-a-full-page-ad-slamming-apples-move-to-protect-user-privacy-it-didnt-go-well.html
    https://www.cnbc.com/2021/04/26/apple-ios-14point5-iphone-update-is-out-now-heres-whats-new.html
    https://www.cnbc.com/2021/02/04/snap-unity-warn-of-impact-from-apple-ios-14-idfa-privacy-changes.html
    https://www.cnbc.com/2020/12/15/apples-seismic-change-to-the-mobile-ad-industry-draws-near.html
  7. Like
    hishnash got a reaction from leadeater in A Mac Pro with RDNA2 Graphics Has Leaked on Geekbench   
    Yes metal has the full support for the same types of ray intersect tests and Vulkan or DX, in-fact some these arrived in metal long before they did in Vulkan/DX. Redshift is one of (a few) of industry leading rendering engines that have added macOS GPU rendering support (using these apis).  


     
  8. Like
    hishnash reacted to leadeater in A Mac Pro with RDNA2 Graphics Has Leaked on Geekbench   
    Edit: Argh nvm, bloody Intel naming. Forgot the Xeon W 3000 series were not LGA2066
     
    Seals the deal for me, fake news (the CPU at least)
  9. Like
    hishnash reacted to leadeater in A Mac Pro with RDNA2 Graphics Has Leaked on Geekbench   
    The entire thing could be a lie, reported GPU model and everything. Mac OS supports RDNA2 so it could be any old AIB 6900XT.
  10. Funny
    hishnash got a reaction from starry in Words aren't enough. We need ACTION.   
    Being made difficult to repair (eg not socketed etc) is nothing at all to do with right to repair. Something that is 100% soldered and very complexly soldered can still be 100% supporting your right to repair it is just harder to repair. The same is true in cars not every person on the side of the street and take and engine apart and put it backtogher without screwing up.
  11. Like
    hishnash reacted to elfensky in Tile bashes Apple’s new AirTag as unfair competition. It will be asking Congress on Wednesday to take a closer look into Apple   
    As I said in a previous post, nothing is stopping you from selling that subscription only on the website. Look at floatplane as a perfect example.
     
    What you can't do is charge less on your site. But you can totally ONLY sell on the site.
  12. Like
    hishnash reacted to Obioban in Tile bashes Apple’s new AirTag as unfair competition. It will be asking Congress on Wednesday to take a closer look into Apple   
    The AirTags are using Apple's U1 chip to get that <10 cm accuracy-- not bluetooth. These should be significantly more accurate than any other tracker out there. 
     
    Not sure what you mean by all iPhone scanning being too much-- it's end to end encrypted, so only the AirTag owner can see the results. Not Apple, not any governments, etc. 
     
    Tile uses subscriptions for non local (bluetooth range) finding things. 
     
    The rest of your complaints seem to just be generic apple ranting-- in the case of AirTags, apple is the cheapest for the feature set, longest battery life, waterproof, serviceable, not claiming they invented the category, privacy minded, and doesn't require a subscription. 
  13. Like
    hishnash reacted to Bombastinator in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    The 20% thing I was already counting on.  Even at 20% slower the thing is still fast enough to game on cpu wise.  Not the very fastest mind you, but fast enough to work OK.  Which is really all that is needed.  The question is will they be able to beat a or even get near a 5700xt using emulation? There are GPUs that are more than 20% faster than a 5700xt.  They’re not integrated though. 
  14. Informative
    hishnash got a reaction from thechinchinsong in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    I expect we will see HBM2e stacks on the pro SoC (that will also be a tile based multi chip solution, apple have a load of patents in this space as to TSMC).
     
    This iMac is just a replacement for the 21" iMac (that for a while now has only supported 16GB).  Apple are still selling the 27" iMac so the message is clear that this is very much not a `professional/prosumer` device. I expect we will either get a May event or they wait until WWDC for the next round that will be the next SoC with different IO (memory and PCIe bandwidth) 
     
    Depending on your use case it is worth nothing that TBDR pipelines used in the M1 do massively reduce the memory bandwidth needs: rather than rendering out each object to a buffer then pulling all those buffers back into the gpu to blend them the screen is split into tiles and each tile is rendered on the GPU, the final output buffer stays on the GPU core (in a small high on die memory) each object is rendered and then depth tested and blended into the final output. This does massively reduce memory bandwidth needs for rendering of 3D scenes it also enables a lot more culling since the accumulated depth buffer is used before rendering the next object to cull fragments that will fail the depth test before the fragment function is called. But in compute compute tasks this is not so relevant (sure having that on core memory helps but you need to code for it very well and that is hard to do in many cases).  But in general i agree we will see HBM2e memory on the next higher end iteration. 
  15. Agree
    hishnash got a reaction from JLO64 in AppleCare+ can now be extended beyond 3 years   
    Louis Rossman does not deal with people who have AppleCare+ coverage as this coverage includes apple just replacing the device, even if you run it over with a truck they will replace it for a resale fee (full device replacement cost is $299 even if you bring it in a top of the line macPro with holes drilled through the case and internals).

    He tends to also not repair device that are that new, if you look at his videos very very few device are ever within 3 years of release. And remember when he does work on a device that is younger most users don't even buy the 3 year coverage.
  16. Agree
    hishnash got a reaction from BuckGup in AppleCare+ can now be extended beyond 3 years   
    is that not what every insurance program provides? (this also extends your warranty as well).
  17. Like
    hishnash reacted to saltycaramel in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    Also, thank you Linus for trolling me (not me specifically) again with the “iPad on a stand” comments about the M1 iMac 🙏🏻 😏

    An “iPad on a stand” with the first and only “larger than 4K” 16:9 display affordable by the masses. And with a full desktop OS, and virtualized Windows. 
     
  18. Like
    hishnash got a reaction from Taf the Ghost in Words aren't enough. We need ACTION.   
    The issue is not parts but what you define as a part, even in the Automotive industry some vendors will define large segments as parts so you need to fully a lot of the car when something failes with most devices these days bing soldered (and there is good reason to do this) do you define a part as an item that can be connected without solder (that is how the Automotive industry works) or do you define a part as something that can be replace on a board, what happens when you have an SoC package (like AMDs Zen cpus) does AMD need to provide chiplets or is it ok just to provide the entire premade package... were do you draw the line. And who provides these parts and at what price, is sony required to make a loss selling AMD SoC to people who want buy the and build their own computers?

     
  19. Agree
    hishnash got a reaction from LAwLz in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    No apple have their own GPU tec, I expect it will be based on apples own inshore GPU cores that have much better perf/W than those from AMD. Likely apple will go with a McM package for the SoC with GPU dies in there, and the dedicated add in card will just be a repate of that package without any CPU dies and more GPU dies but otherwise the same as the main SoC.

     
  20. Agree
    hishnash got a reaction from LAwLz in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    I expect we will see HBM2e stacks on the pro SoC (that will also be a tile based multi chip solution, apple have a load of patents in this space as to TSMC).
     
    This iMac is just a replacement for the 21" iMac (that for a while now has only supported 16GB).  Apple are still selling the 27" iMac so the message is clear that this is very much not a `professional/prosumer` device. I expect we will either get a May event or they wait until WWDC for the next round that will be the next SoC with different IO (memory and PCIe bandwidth) 
     
    Depending on your use case it is worth nothing that TBDR pipelines used in the M1 do massively reduce the memory bandwidth needs: rather than rendering out each object to a buffer then pulling all those buffers back into the gpu to blend them the screen is split into tiles and each tile is rendered on the GPU, the final output buffer stays on the GPU core (in a small high on die memory) each object is rendered and then depth tested and blended into the final output. This does massively reduce memory bandwidth needs for rendering of 3D scenes it also enables a lot more culling since the accumulated depth buffer is used before rendering the next object to cull fragments that will fail the depth test before the fragment function is called. But in compute compute tasks this is not so relevant (sure having that on core memory helps but you need to code for it very well and that is hard to do in many cases).  But in general i agree we will see HBM2e memory on the next higher end iteration. 
  21. Funny
    hishnash got a reaction from kirashi in Words aren't enough. We need ACTION.   
    For sure but if you want some law to be passed for Right to Repair trying to also at the same time re-write IP and Copywrite law around the world that might take many many may more years to pass if ever. 
     
     
    Im not sure that many go out of their way (put extra work in) to make it harder, but rather they do not consider these things at all when building new products. Things like requiring that you calibrate the sensors when connecting them to the SoC, that is a valid thing to do, it saves on having calibration stored on the sensor making the sensor less complex, saving money and likely power. The issue is they do not think about the repair market and do not provide (or even document) tools/interfaces needed for others to do this same calibration.  

    What is needed is some legislation that requires that all the interfaces between components is documented. That minimised the private IP that needs to be exposed, but still enables people to find/build solutions that can replace broken parts. And unlike demanding they sell repair parts it enables old devices to be repaired.

    You can't really demand a manufacture sell spare parts (at the low volumes they will be) for 10+ years after they stop shipping a product as the cost of keeping production lines open for such low volume part production would be a massive risk. Much better if they are forced to provide the documentation that would let others (if there is demand) find/developer replacement chips of their own. maybe you could have a thing saying if your not able to provide the parts you need to provide this documentation but that risks companies going out of bis without every sharing that spec (think of all the homeAutomation east that is already out there due to startups that have flopped) 
     
    The only part that apple does this with is the touch id and face id sensors and that is not a ID they are pinning but a public certificate for validating the encryption of the data it is sending ( the reason for this is complying with laws around apple pay, apple put in the extra dev work to rather just disable apple pay but that is putting in extra dev work so not likely to happen). Other parts (like the camera not working properly) are not due to ID issues but rather calibration issues, unlike older camera modules the cameras in the latest iPhones do not store their calibration information on the camera but it is instead stored in the SoC. Without this info you can `use` the camera but the raw bytes that a read from it are rather useless, not the camara api can be used but randomly crashes as soon as it hits data it can't interpret, this is clearly not an ID based check. What should happen is they should be forced to document this calibration process, so that someone else can build the tools needed or provide the software tools to let you do it yourselves. 
     
    It is not that they are making deals to not sell it is that they are not making deals to enable the sale. 
     
    When you ask a factory to do a custom order you pay for tooling, design etc if the factory, you pay before they even start to make anything. If your a large vendor (like apple, Samsung etc) your tooling costs might be very large since the factory might need to tool up multiple concurrent production lines just for your order, the per unit cost after that work has been done is low. If the factory were to turn around after you have paid for them to build up all this tooling and just sold these chips to other people they would need to create an explicit contract with you to permit them to do this.
     
    Most vendors have really only done this when it involves claims of false representation, eg using third party parts but claiming they are original. I think under right to repair laws the punishment for doing this should be increased it should be explicitly mentioned in the laws that if you operate a repair service you must be transparent about the origin of the parts you are using. 
  22. Funny
    hishnash got a reaction from kirashi in Words aren't enough. We need ACTION.   
    Being made difficult to repair (eg not socketed etc) is nothing at all to do with right to repair. Something that is 100% soldered and very complexly soldered can still be 100% supporting your right to repair it is just harder to repair. The same is true in cars not every person on the side of the street and take and engine apart and put it backtogher without screwing up.
  23. Agree
    hishnash got a reaction from Distinctly Average in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    I expect we will see HBM2e stacks on the pro SoC (that will also be a tile based multi chip solution, apple have a load of patents in this space as to TSMC).
     
    This iMac is just a replacement for the 21" iMac (that for a while now has only supported 16GB).  Apple are still selling the 27" iMac so the message is clear that this is very much not a `professional/prosumer` device. I expect we will either get a May event or they wait until WWDC for the next round that will be the next SoC with different IO (memory and PCIe bandwidth) 
     
    Depending on your use case it is worth nothing that TBDR pipelines used in the M1 do massively reduce the memory bandwidth needs: rather than rendering out each object to a buffer then pulling all those buffers back into the gpu to blend them the screen is split into tiles and each tile is rendered on the GPU, the final output buffer stays on the GPU core (in a small high on die memory) each object is rendered and then depth tested and blended into the final output. This does massively reduce memory bandwidth needs for rendering of 3D scenes it also enables a lot more culling since the accumulated depth buffer is used before rendering the next object to cull fragments that will fail the depth test before the fragment function is called. But in compute compute tasks this is not so relevant (sure having that on core memory helps but you need to code for it very well and that is hard to do in many cases).  But in general i agree we will see HBM2e memory on the next higher end iteration. 
  24. Like
    hishnash reacted to leadeater in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    Actually it was, though I suspect you aren't a regular WAN show viewer but it's been talked extensively on there as well as promotion of the funding campaign for it too. It's a fairly regular topic on WAN show, and if not a topic talked about within topics a lot.
  25. Informative
    hishnash got a reaction from leadeater in Apple April 20 event - New iMac and iPad Pro with M1 chips, 'AirTags' and a new iPhone color   
    you mean running Vulkan or DX, VM GPU performance (for DX) is ok on the m1 (you are loosing about 20% of perf compared to running native metal) (the DX driver is written by the VM provider that maps the DX into Metal instructions). 
×