Sorry iOS, I'm switching to Android
- Thread starter Fishfam4
- Start date
I'm factually correct. I suppose I'll educate you as well. The biggest gain you'll see from going from a 32-bit to a 64-bit processor is in memory addressability. The iPhone comes with 2GB of RAM that you're unable to upgrade. You can address 2GB of RAM with the 32-bit processor.
It has nothing to do with Android or Apple. You're letting your choices blur your vision. I'm sorry you chose to join the condescending behavior when you know nothing about the topic. Next time, don't be fooled by marketing gimmicks and you won't feel the need to defend your emotional responses the marketing. You certainly weren't acting rationally when you believed the 64-bit processor meant much of anything nor were you when you made this post.
Thin air? You suggested you are ignorant until you work for a living. You were entirely ignorant in your post. By your logic, that means you haven't worked for a living.
The truth doesn't upset me. Your ignorance does. I explained to someone else what the biggest benefit of the 64-bit processor is. It's not utilized in the iPhone 5s and cannot be utilized. As such, the "superior processing" isn't superior. I'll try to explain it in a way you'll understand. If you have a car that CAN travel 100mph, but you put a limiter on it to prevent it from going faster than 50mph, you can't reasonably claim it's faster than a car that travels 50mph. The hardware decisions Apple made mitigate the potential benefit of the 64-bit processor. Educate yourself and stop being so easy for marketing.
Let's work through that post in pieces. We'll start with the easiest first. Most people don't know what 3G and 4G are either. But, marketing certainly uses that and takes advantage of people believing a larger number has meaning. Are you aware marketing teams went to 3GPP begging to get the 4G standard changed so they could call LTE 4G even though it fails to meet 4G specs? Purchasers don't NEED to know what it means. Marketing works better when they don't know what it means. That's why marketing was quick to use the larger number to suggest a superior product.
The harder of your two questions relates to benchmarks. I'm going to point something out to you that you've apparently missed. The benchmark tests didn't change with the 5s. Apple devices have always performed better with weaker hardware than their competing devices. With that in mind, you shouldn't assume it's the 64-bit processor making the difference. The difference existed prior to the 64-bit processor. It's asinine to believe the processor makes the difference when the difference existed prior to the usage of the processor. The difference is the software and hardware were built to work together. iOS is optimized to Apples devices. Android is meant to work on a wide range of devices. There's a concept that extends beyond electronics that suggests the more things you do, the less you do well. You can't truly focus on being strong at any given task when your focus is spread over a larger number of tasks. Samsung is looking into releasing their own OS so they can reap the benefits of software optimized to work on their devices.
Yes. I will say it's a marketing gimmick. You're not getting a great deal of op-codes out of the 64-bit processor to improve performance. The addressability of memory isn't being utilized in a device with 2GB of RAM. Would you like to spend a few minutes to research on how a 64-bit processor improves performance over its 32-bit counterpart before you make another silly post?
the_tech_eater
Well-known member
I have researched it. It's you that needs to research before you make another silly post! Yes 64-bit works best with 4GB+ ram, but it's basically like this: if you had 32 people to work for you, they couldn't get it done as fast as 64. Thats exactly how it works with 32vs64bit. Add more floating point registers that are in the A7, and it's the most powerful chip on the market!
http://www.trustedreviews.com/opinions/what-is-64-bit-iphone-5s-a7-chip-explained
No. It's not. You're confusing processing speed versus the processing resolution. If you have a 2GHz processor, that processes at twice the speed of a 1GHz processor. That's the analogy you gave with 32 people instead of 64.
Most of that article was poorly written. Long-term, the move to 64 bit does make sense. Eventually, they'll be using the appropriate hardware to leverage the device. Let's give you an analogy, as you seem to like those. Eventually, it makes sense to get married. That doesn't mean it makes sense to propose to the girl on the first date. Adding the 64-bit processor prior to including the hardware to take advantage of the architecture is no different. You're not taking advantage of the single most beneficial aspect of the processor. The rest of the gains are marginal.
Don't attempt to tell me I'm being silly when you don't understand that processing speed is what makes an opcode complete faster.
the_tech_eater
Well-known member
You are being silly. And quite frankly I don't want to argue with you, because that would be bringing myself down to your level.
After you made the mistake of telling me 64-bit processors make the change that increasing the processing speed makes, I'd take a cheap shot and leave the discussion too. I'm being "silly" by understanding the topic. Nobody makes the claim that a 64-bit processor handles the task at twice the rate of its 32-bit counterpart other than you. With such a weak foundation of knowledge, I wouldn't want to argue if I were you either.
You can't bring yourself down to my level. You don't know enough about the topic to do so. You've used inaccurate analogies to attempt to explain a concept you don't understand to someone with an engineering degree. The one article you linked used some bad ideas along with some good information. You aren't educated on the topic enough to know which pieces are which. Adding more hardware simply because we will eventually make use of that hardware isn't a good decision making process for the CURRENT product. We've agreed there is a maximum RAM that a processor can use. Would it make sense to add more RAM to the 32-bit processor because eventually we'll be using that much RAM with a 64-bit processor even though the device wouldn't be utilizing the increase in RAM? Of course not. Yet, you're telling me I'm being silly for pointing out this decision making process doesn't make sense.
Speed of a processor isn't measured in bits. It's measured in Hz. A 1GHz processor completes an cycle in 1 billionth of a second. A 2GHz processor completes the same cycle in half the time. This translates to your analogy. Twice the people can perform the task in half the time. Twice the processor speed can perform the same task in half the time. That same comparison simply doesn't translate between 32-bit and 64-bit processing. You confused what you read to mean something it didn't. Yes, you can combine calculations by doing twice as many calculations at once. But, that makes sense in a pipeline. If the calculations aren't pipelined, you're not gaining the performance increase you believe you're gaining. Pipelining typically takes place for portions of code, not the entire code. You'll see a performance increase, but only in a some portions of the code. You have more room for opcodes that can be used to optimize processes and reduce the number of cycles required for each opcode. For example, there are multiply opcodes that can perform at better rates than others depending on the processor. Some may take 1 cycle, some may take 2, some may take 10. This can show performance gain.
But, the greatest gain STILL comes from the increased addressability.
Don't argue. It simply doesn't make sense for you to do so. In a technical conversation, you're in a place to learn rather than argue. Take advantage of the free education you were just given. Don't begin to act like you deserve to be condescending until you're able to analyze the articles you're linking and understand what points are misleading and which have merit.
HAWK
Well-known member
That old galaxy charger works in the new galaxy just the new charger takes up more room. You can still use standard micro USB on the right hand side to charge it.
the_tech_eater
Well-known member
If clock speed is what matters, than why is the A7 faster than processors with more than double the clock speed? And also, the A7 is only .2 ghz faster than that A6 and it benchmarks, and in a lot of cases performs better than the A6. Care you explain that smarty?
HAWK
Well-known member
You're confusing two ideas. There's obviously more to a processor than purely clock speed. I was explaining from within the context of your analogy. If you don't understand the difference between clock speed and bits, there's no point getting into more complex systems. I'm not attempting to suggest clock cycle is the only measure of a processor's efficiency. I'm just not overstating the value of a 64-bit processor compared to its 32-bit counterpart.
If everything else remains equal, doubling a processor's clock speed doubles its ability to compute data. If you take a 1GHz processor and the only change is to increase it to a 2GHz processor, tasks take half the time.
If everything else remains equal, doubling the bits from 32 to 64 increases the processing speed of some calculations while providing absolutely no benefit to others. Take the idea of a Multiply/Accumulate calculation. This is something like y = a(x+z). If the processor is built to include an opcode just for the MAC, it's possible to make this a single cycle calculation. If not, it's going to take however long the processor requires to process both the add and the multiply. It doesn't matter if the processor is 32 or 64 bits. Why? Until the x+z is added, the result can't be multiplied. Sure, you can do more addition functions at a time by utilizing the larger address space. But, you wouldn't in this case. Practically, it's not possible to pipeline every instruction you give in such a way as to gain twice the processing power from the extended size as you suggest.
The three most common ways to leverage a 64-bit device to gain processing power are: more space for streamlined op-codes, more simultaneous calculations, more RAM addressability.
We've already addressed simultaneous calculations to some extent. This is very specific to the application being used to benchmark. In some, it may appear to be a huge difference. In others, you're unlikely to see a change. Different applications have different degrees of pipelining used in their design. As an example of pipelining, let's say you have a collection of savings account information and it's time to compute interest. Every account needs to be multiplied by 1.03. If the multiply takes two cycles, you can start computing on the first account during the first cycle. After the first cycle's work is finished, you can finish it in the second cycle while doing the first cycle's work on the second account in the second cycle. This organization means you can do 1,000,000 accounts in 1,000,001 cycles instead of 2,000,000. Depending on the registers being used to do this, it's possible to leverage the 64-bit processor to decrease the time. But, that's not keeping all things the same. That's making additional hardware decisions. It's likely these decisions would be included. But, they'd STILL only see a benefit in these types of cases. When you're not doing a large number of similar calculations immediately following each other, you're not going to see the benefit.
In terms of op-codes, let's look at the most basic architecture. We'll use an 8-bit architecture and we can expand the idea from there. With each address being a byte in size, we have 8 possible positions to include all of the information the processor needs to handle the instruction. Let's use 3 of those bits for op-codes and the remaining 5 for register addressing. This is an arbitrary decision. You can choose however you'd like. With only 8-bits, we're severely restricted on space. 3-bits for op-codes mean we can handle 8 different instructions. That's not a very advanced system. With 5-bits for addressing, we'd likely split the space into 3 sections: 1 bit to determine which register to save the result, 2-bits for each register to use in the calculation. With only 8 possible commands, we likely don't have room for add, subtract, multiply, and divide. That would use half of our available commands. If we also include save/load, we're at 6. This means we can't include and, or, and not. We could get away with just using AND and NOT, but that's it. We also only have one type of save/load, so it's not very efficient. It's easy to see how the 8-bit space restricts us. If we expand to 16-bits, we can extend the op-code space to 4 bits, double our op-codes, and increase the efficiency of our program by opening up a great deal more possibilities for placement of data. We can further extend this to 32-bits. At this point, we see a lot of architectures using the idea of byte addressability. At this point, we consider the value of having EVERY instruction maintaining the same length. If I want to OR two locations, I need the op-code space, the two locations, and a place to store the result. That could require more bits to explain than something like an unconditional branch. The branch just needs to say "go to this address no matter what." It's the op-code and an offset. This style allows you to pack more instructions into the same space. At this point, we already have a large number of op-codes available. While the 64-bit processor can easily spare a single bit to double the op-codes, this also requires building streamlined operations for each. That's a HUGE investment in research and development for negligible performance improvement. It's not practical at this degree.
Most users understand that more RAM offers faster potential processing. The reason it offers quicker processing relates to the time it takes to get information. Let's use the analogy of cooking preparation. You're going to make beef stew. You'll need to cut some beef, carrots, potatoes, and celery. You'll also need to mix spices and other ingredients into the pot. It takes less time to cut the carrots if they're already on the table in front of you than if you have to walk across the kitchen to get them out of the fridge. RAM works a lot like this. It keeps the data closer to the processor. As a result, it speeds up computation. More bits means you can have more ingredients on the table, but only if you have enough table space to accommodate the added ingredients. Going from 32-bit to 64-bit without increasing your RAM completely ignores this benefit. You can have more ingredients, but your table is the same size.
Many processors use these ideas to specialize. A processor used primarily for Digital Signal Processing will want to include measures to make the FFT faster to compute. It makes sense to set aside registers, create butterfly architecture, and use op-codes to reduce the time it takes to compute each butterfly. If the end user isn't likely to be building these types of applications, they may not use the FFT as often. As such, it doesn't make sense to devote this much attention to the calculation. We could benchmark two processors side by side and use the FFT as our application and find one processor to be far superior to the other. But, that doesn't mean it processes that much faster than the other. There are countless decisions like this one that go into designing a processor. If I were to make the B1 processor for DSP and skip the FFT streamlining, I could make a HUGE improvement to the B2 by adding this streamlining. This is true even if I don't change the bits, clock speed, or anything else. You can't simply ask "why does this processor have a faster clock rate and perform better?" The answer isn't that simple.
It's adorable you're STILL acting condescending when it's beyond apparent you have absolutely NO clue what the discussion is about. Do you have any more questions that you'd like to ask with a condescending tone while simultaneously highlighting your ignorance in this topic?
fatema2014
New member
ajl
Well-known member
anony_mouse
Well-known member
The A7 is certainly a very powerful chip, but I'm sorry to say that you certainly do not understand 64 bit computing.
FFR
Well-known member
anony_mouse
Well-known member
Thank you for the friendly post. I'm afraid I am right. 64 bit computing is faster, when there is more than 32 bits of data to process, or more than 32 bits of memory to address. When there isn't, it may actually be slower as there's more data to be moved around the system. Benchmarks from the introduction of amd64 processors in PCs are well worth looking at to get an idea of the likely impact.
It's *probably* a good idea for smart phones to slowly move to 64 bit architectures, but it will not be a dramatic improvement in most cases. The A7 is a very fast chip, but that is primarily because it uses a very fast new ARM core, not because it uses a 64 bit architecture.
This is an interesting area. Try reading some books on CPU architecture if you're interested in it.
Last edited:
taz323
Moderator
Friendly reminder folks.........
P.U.P.P.A.H
(Be Polite, Understanding, Patient, Professional, Attentive and Helpful)
Be polite. We ask that you keep your language clean and polite.
Be understanding and patient. You may be a power user or developer, but the person you’re talking to on the other end may be a complete newbie. Be patient, and provide all details possible if you’re posting about a problem (descriptive thread title always helps).
Be a pro. Don’t get into arguments over dumb things. If someone responds in a poor or rude fashion, ignore it. If you notice a pattern of rude or poor behavior from a particular member, report them to our moderators. If you don’t have anything nice to say, don’t say anything at all. In a nutshell, ‘think before you post’.
Be attentive and helpful. Answer questions in a timely manner when you can, and provide thoughts and suggestions that might lead to the solution of a members’ problem.
P.U.P.P.A.H
(Be Polite, Understanding, Patient, Professional, Attentive and Helpful)
Be polite. We ask that you keep your language clean and polite.
Be understanding and patient. You may be a power user or developer, but the person you’re talking to on the other end may be a complete newbie. Be patient, and provide all details possible if you’re posting about a problem (descriptive thread title always helps).
Be a pro. Don’t get into arguments over dumb things. If someone responds in a poor or rude fashion, ignore it. If you notice a pattern of rude or poor behavior from a particular member, report them to our moderators. If you don’t have anything nice to say, don’t say anything at all. In a nutshell, ‘think before you post’.
Be attentive and helpful. Answer questions in a timely manner when you can, and provide thoughts and suggestions that might lead to the solution of a members’ problem.
FFR
Well-known member
So your trying to compare an and x86 processor with a 28 nm ARM v8 utilizing a Cyclone micro-architecture,
lol, ok you go do that.
The difference is Apple introduced the 64bit ecosystem to further their 64bit agenda, unlike amd which built the football stadium without a team or any fans.
btw I was being very friendly, unless your really Anand Chandrasekher :laughing: :beer:
Last edited:
Similar threads
- Question
Trending Posts
-
-
The iMore 20K / 50K Post Challenge - Are you up for it?- Started by Jaguarr40
- Replies: 31K
-
-
-
Facebook
Twitter
Instagram