Intel i5 vs AMD Quad-Core A10-4600M Accelerated Processor

I will be an engineering student this fall, and it'll involve using quite some 3D modelling software. I'm trying to choose between 2 HP laptops, one has the AMD and the other the i5. However, for the intel it says there's up to 1696MB total graphics memory, and the AMD one has up to 3060MB. So my question is will the AMD processor be significantly worse than the i5? Or will it still be good enough? Btw the graphics for the AMD is Radeon HD 7660G and the intel is Intel HD Graphics 4000.

I'm in favor of the AMD one, but the fact that it is an AMD processor makes me more reluctant, because the intel i5 is better to my knowledge, so please help me out!

Just comparing the 2 processors and GPU's is good too!

 

Well, let's look at benchmarks. While it's nonspecific, the Passmark CPU list is good to get an overall picture of performance. I'm not sure which i5 you're looking at, so I've chosen the fastest one, the 3360m.

i5 3360m: 4527 points
A10 4600m: 5129 points

The AMD is slightly faster, moreso if you're talking about a lower-spec i5.

And for the GPU, I'll be using NotebookCheck's 3dmark11 benchmarks, again to measure overall performance.

Intel HD4000: Avg 638.9, max 802.
AMD7660G: Avg 1142.5, max 1150.


The A10 wins by a little in terms of CPU speed, and a more significant margin in terms of GPU speed. Also, if you're going to be doing anything that uses OpenCL, I'm pretty sure the AMD will have better drivers as well. It should be able to pull off gaming at 720p on lower settings as well.

If any of your software will be using CUDA, you might want to consider at least a low-end dedicated nVidia GPU.

 

I am finding above post hard to believe. Anandtech review had it losing even to i5-2410m which is the slowest 2nd generation i5 CPU. in fact its even losing to ULV i7 sandy bridge cpu.

AnandTech - The AMD Trinity Review (A10-4600M): A New Hope

 

Passmark is a very inaccurate benchmark. According to Passmark, the i7-3615qm is faster than the desktop i5-3570k.

The i5 will be faster for CPU heavy tasks (e.g. photo/video editing) and the A10 will be faster for GPU heavy tasks (e.g. gaming). For 3d modelling the i5 will probably be a better choice, though really if you're a freshman engineering student either you'll do pretty much all of your work on your school's computers or the work that you'll do won't be intensive at all.

But if you really need the best performance possible, I would get an i7.

 

If you would like to see nice big numbers in CPU synthetic benchmark programs than Intel i5, otherwise A10-4600M with 4 cores and Radeon 7660G is the winner for 3D modelling, games, running more programs parallel.

 

PassMark is a random number generator. I really wish people would take a look at what kind of comparisons it gives between "known" processors before quoting it.

What is the name of the 3D modelling software? This is important because which processor is better depends on whether your software is GPU-accelerated and to what degree. The 4600M has the faster GPU, but the slower CPU so if your software gets a significant benefit from the GPU, it's probably the way to go. If not, then the i5 is better choice because, despite the "quad core" label, the 4600M will be slower in both single-threaded tasks (by a huge amount) and multi-threaded tasks (by a moderate amount).

 

I would buy the A10 assuming that the battery life isn't horrible. It has much better GPU performance and more cores for multi-threaded CPU performance. The only place it loses is single and dual threaded performance.

It's basically better single thread performance and battery life vs GPU and extra cores.

P.S. the amount of quoted system memory the IGP can use is worthless as a benchmark, they aren't powerful enough to need more than 1GB and probably not even that.

 

..intel graphics as well as apus can use any amount of the ddr ram on the system. But some manufacturers set the ram-size in the bios for the intel graphics for some reason.. Say, they have one bios for all their skus, and one setup comes with 2Gb ram - then you need to limit it. If you can avoid problems with 1Gb graphics card setups by limiting it as well - then you would do that. And there's suddenly no way to get around it.

The apu setups control that via software instead. Same with the clock-speeds, which are stuck in the same way on the intel graphics (at annoyingly high frequencies, much higher than needed for just desktop work, for example). So that part of the apu/radeon driver is definitely good. So is the performance, specially when thinking about performance per watt, and so on. Switching between igp and dedicated graphics is also not exactly flawless since the ram-areas you address are physically different. So there's problems here if you expect to switch between battery and plug, or different profiles, etc..

Where you will regret buying a radeon card, is when you start trying to configure the drivers. Then you're suddenly in all kinds of trouble. They just look... amateurish, and don't give you very good or easy options. Same if you use hdmi out. It's not accurate enough for office-use -- anything outside presentations or 3d frame contexts, and you start to squint very quickly..

OpenCL performance.. The apus win by ridiculously massive margins since the gpu cores can access working ram as fast as the processor.. The quad-core runs on all the apus so far are insane for a commercial PC. So if you expect to be using programs that have opencl support, the apus win. Even if for example a flash-container with a "hardware-acceleratable" format won't actually be decoded more efficiently, of course. So until habits evolve a bit, you're only going to see the increased performance on specific programs that support or use OpenCL.

At the end of the day, though -- agree with the comment above about battery-life. Most important thing, imo.

So if you can have "good enough" performance (and more than that) on something that has the best power-use. Then why pick something that has slightly higher peak performance on x86 number-crunching - if you don't actually need it..?

 

Nope. The i5 will crush it in multi-threaded performace -- not by as much as in single-threaded, but still by a lot.

AMD pulled a brilliant marketing trick by putting 4 integer cores in there and calling it "quad-core" based on that. It is not really a quad-core processor: a whole lot of important things are shared between pairs of cores ("modules"): instruction fetch, branch prediction, FPU, L1 instruction cache, L2 cache... the only reason to call it a "quad-core" is that the integer units and L1 data cache are not shared. The result is somewhere in between having 4 real cores and having 2 cores with hyperthreading (but since Intel's cores are so much faster to begin with, 2 hyperthreaded cores win in practically everything).

 

The trinity chips still outperform the "true" quad-cores from the previous iteration, though.. At lower power-use.

 

It doesn't. The A10 has the multithreaded performance somewhere around an Intel i3.

Stock, maybe. But Llano can be overclocked while Trinity is locked down. Any Llano quad at a conservative 2.4ghz OC will beat the A10 multithreaded.

 

The scientific explanations are might good to confuse people, but end-users do not cares what is under the hood, especially if ain't change anything on the facts. From experience and from other Trinity users the 4 integer cores are just simply better to run more than two programs parallel than 2cores+HT. And if you are remember AMD had a great demo to demolish the power of 4integer cores +IGP, while even the I7 Intel failed to show anything similar like that.

 

What programs? My work involves some of the most CPU-intensive applications out there and we routinely look at comparisons of CPU architectures as they are released (more for servers than laptop, but since everyone has a laptop anyway and the architectures are extremely similar, the latter serve as advanced notice). Trinity is based on Piledriver which is a minor revision of Bulldozer. There are very, very few programs in which a Piledriver module (i.e. pair of integer cores + associated resources) beats a hyperthreaded Sandy Bridge core (never mind Ivy).

Of course, most applications which are run in parallel (web browser, email client, office) are bottlenecked by storage rather than the CPU so you can certainly get better performance from an SSD (or a faster hard drive) even if you are using an inferior CPU, but this is not what multi-threaded performance is about.

 

Not sure what kind of "most CPU-intensive applications" are you meaning, maybe not important for 99% of the folks. If you look back the links in my previous post, it shows a situation which can be useful for consumers; one core work with movie encoding, second core play movie, 3rd core with igp play game, 4th core play another movie and the laptop still alive. This is a multithreaded situation where Intel i5 have no chance.

 

A10.

just go dual channel RAM

 

Every single time this topic is brought up you seem bent on ignoring all the actual benchmarks for some reason. In terms of CPU performance, Trinity is slower 99% of the time than even Intel's last gen i5, let alone the current gen. Anandtech shows this, Tom's Hardware shows this, Hot Hardware shows this. Even the benchmarks that you yourself posted in this thread show that an a fast Intel dual core would be faster than a slower AMD quad.

How long are you going to ignore the truth and mislead other people?

 

If laptops, I'd buy only Trinity. Cheap, fast, cool, do everything what Intel and more and longer!

 

..Going by tomshardware numbers on their test, I doubt that's actually true.

Also doubt the frequency scaling isn't going to be possible to tamper with eventually.

 

What numbers? And what do you mean by "frequency scaling"?

 

I assume that he means you will eventually be able to overclock Trinity just like you can overclock Llano.

 

I'm talking about multitasking capabilities, GPU performance, power consumption and you are coming again with your synthetic benchmark links, claiming I'm ignoring the truth, mislead people. You know what; Yes I'm absolutely ignoring Chinebench, PCMARK Vantage and other synthetic CPU craaps. However I ain't mislead people with linking real world situations! The fact is; for overall AMD has better mobile products than Intel and I'm sorry if this hurts you.

 

Trinity overclocking automatically according to TDP limit, temperatures and loading situations, while the old Llano is an un-optimized APU, which need manual overclock, undervolting.

 

Cinebench is just a benchmark of Cinema4D, which is a "real world" program. It's highly threaded rendering program. Which is what OP was talking about. If a CPU is slower in Cinebench, it's going to be slower in Cinema4D too.

And what about the Photoshop benchmark? Photoshop is synthetic now? LAME? Autodesk? Think about it, when pretty much every CPU benchmark, synthetic or non-synthetic, single threaded or multithreaded, shows that the i5 is faster, I think it's pretty safe to assume that the i5 is faster, no?

On what basis do you say that 4 Trinity cores are faster than 2 Ivy Bridge cores for 3d modelling? Because you want it to be?

When all the 3d modelling benchmarks like Cinebench and Autodesk show that the i5 is faster, you're going to have to do better than linking 2 unrelated youtube videos and sprinkling in some hype on top.

That's your opinion then I have no problem with that. However, some of the "facts" you talk about are just plain wrong. Like in this thread with media encoding. You can have whatever opinions you want, if you're going to post things that are just untrue, then all you're doing is misleading other people.

AMD generally has a better iGPU while Intel generally has a better CPU. "Better" depends entirely on the needs of the individual.

 

Doubt it. Locked SB isn't OCable (in a meaningful way, see next paragraph), locked IVB isn't OCable, locked desktop Llano isn't OCable. Locked multipliers are locked multipliers, there's no evidence that Trinity is going to be any different.

The other way to OC with is through BCLK which gets you like 100-200mhz and makes your system unstable. And even that's impossible for most people because almost no recent notebook has the BCLK exposed (the only one I can think of is the MSi 16f2 with an unlocked BIOS).

 

Probably. But we are not talking about the general population -- almost any CPU/GPU combination is good enough for them.

First, I don't understand why anybody would be playing a game and watching not one, but two movies at the same time on the same laptop -- this is not a real world scenario by any means. However leaving that aside, the i5 most definitely can handle such a workload. All Ivy Bridge processors have Quick Sync which is by far the fastest way to encode movies (faster even than desktop GPUs). Playing two movies at the same time should not cause any problems. The only question is which game you want to run -- any casual game is fine, AAA 3D games not so much (but Trinity would likewise choke on AAA 3D game under these circumstances).

Yes, "real world situations" like watching two movies and playing a game all at the same time on one laptop. Do you own stock in AMD or something? Every reviewer out there agrees that Intel is the best for CPU performance and AMD is best for GPU performance. You are literally the only person on any site I have seen (except for the occasional PassMark enthusiast) who believe that Trinity can compete with Ivy in CPU tasks, benchmark or real application.

Yes, but Llano is fully unlocked (there are no built-in limits to how far Llano can be overclocked; it's restricted only by heat and power draw) whereas Trinity's max Turbo is currently as high as it can go.

 

..um.. well.. it is locked, it's just that the processor states are software controlled. So it has been possible to control the pre-set states with a hack.

Frequency scaling on intel chips is more difficult since it's possible to lock the bus-speeds in bios. Which of course is something retailers and manufacturers prefer, since overclocking is evil. And they much rather prefer to let the laptop manufacturers cripple the bios with their incompetence instead of ours.

..No, it just can't be overclocked with k10stat. Different.

 

you would be using some 3D stuff.

- while intel would win in CPU intense situation, their iGPU absolutely suck.
- while AMD would be slower somewhat in the CPU stuff, their GPUs are quite useable for variety of situations, from gaming to pro software use, so overall you're covered in any situation.

while you would be able to run everything on the AMD, there will be case where you'll see "nVidia/AMD needed to run this" type of situation in the Intel case, and there's nothing you could do about that.

you know, processor speed is what people use to advertize and sell computers. Of course most of us know that this is not even half of what it takes to make a computer run decent, lol. I'm sure if I were to bring my grandmother to bestbuy and let the sales person talk to her, she'll end up with intel due to the faster processor .. LOL.

also, look to see if you could replace the AMD chip later. That would be awesome.

P.S.
I've tried AutoCad on my intel GPU laptop (see my sig), and oh man come to see what slow computer is ... LOL. Same program runs on my nVidia laptop quite well. Thus my point that having nicer GPU is quite important.

EDIT: I dont actually need more CPU, but I'd really really like to have more GPU And I'm on T9500 here (~18 Gflops)

 

Can we get more information about the software you plan on using OP?

This is essentially going to come down to two factors. Which do you favor more for your workloads, cpu or gpu performance?


Going by what the OP has said, I would recommend the AMD APU. What was only mentioned was the processor(s) themselves & not any additional discrete gpus. It seems you are more concerned with graphics due to the fact that you will be using 3D modeling software. AMD's superior gpu architecture (but inferior cpu), in comparison to Intel's, will be better suited for the type of workload you stated. We have already established that Intel has the superior cpu but inferior gpu by the benchmarks shown. With the AMD solution you will get an APU that will take care of your task & at a lower price point too.



Sent from my SPH-M580 using Tapatalk

 

This is not true anymore. You would be correct if we were discussing 4 year old laptops -- indeed the Intel 965 in your signature is way, way worse than any AMD or Nvidia card of its time (never mind today). However, Intel changed their strategy a few years back and started moving towards decent GPUs.

Arrandale was their first attempt and while it was not close to contemporary IGPs, it was at least edging towards compatibility. Sandy Bridge was next and, at the time or release, it was actually better than any other IGP out there (though of course Llano was released a few months later and that was 60-80% faster). The current iteration, Ivy Bridge, is only about 20% slower than Trinity (scroll down to the comparison across 15 different games on the bottom of the page).

With that kind of advantage, it really depends on the application. Even something has "3D" in the title, it is entirely possible that it is not sufficiently parallel for the better GPU to make up for the worse CPU.

 

what are we talking about than? R3d claming I’m confusing people about Intel vs. AMD battle, while you just admit for 99% of people ain’t important the “most CPU demanding scene”. The problem is not the 1% (special) people will dial this topic from google (because they already know what is the best CPU for their special CPU program), but a great percentage from 99% who curios which computer is faster? They will end up buying Intel, while their needs absolutly covered by AMD, they could even play near decent quality with AMD and could save lot of battery. So I’m asking now R3d who confusing whom?

The demo is about running more program together and people generally running more programs than one in the same time. They won’t run Chinebench and special CPU programs, trust me.

A10 Trinity base clock is 2.3GHz, while usually autoverclock to 2.7GHz. There is another 500MHz room until 3.2GHz. You could not even reach that with manual way if CPU is unlocked! Same thing with A8-3550MX base clock 2.0GHz, Turbo 2.7GHz. And I cannot reach 2.7 GHz even with manual overclocking, because throttles back, therefore I’m running it on 2.4GHz.

Those test were run near low or medium quality without AA and Anisotropic filtering, while you can turn these details to high on AMD whit loosing minimal FPS. Once you give job to the GPU part, the difference won’t be only 20%, but 40%-50%. The HD4000 cannot even get close to the old 6620G!

For R3D, I just suggest buy an Intel and play with Chinebench/Chinema 4d!

 

I am talking about the needs of the person who started the thread (a prospective engineering student). An engineer may need more out of a laptop than most of the population and it's not obvious from the original post whether the CPU or the GPU is more important.

Yeah, but nobody will run two movies and a game at the same time either. It's just as artificial as Cinebench (probably even more so given that Cinebench at least tells you how well you will run a real program which some people do use). That demo has no more "real world application use" than any other benchmark.

Do you have a link for this? The only way it's possible for Trinity to lose minimum FPS when changing settings is if it is CPU-bound in these games which I rather doubt. In any case, the CPU advantage of Ivy i5's in single-threaded tasks is more than 40-50% and it's close to that in multi-threaded ones.

 

This thread ain't private and the person "abcd3f3" who started did not have a single comment here, I guess he won't even check back to see the discussion. Therefore this thread is for the general publicity. What would be their conclusion, if I am not interfering and approach things other way?

Laptop can be attached to an external tv and play HD movie for the girlfriend, kid, grandma or so, while other person can having fun on the laptop with running more programs, play a game.

Tests coming soon, where Trinity play demanding games in high and ultra whit good frame-rates. All the detail settings will be marked, therefore everybody will be able to compare the results with their own laptops and see how not well perform HD4000 against it.

 

You can that if you want, but that is different from what you said before. My point is, stop posting things that are factually incorrect! I have no idea how I can be more clear than that. You are confusing people befause you are posting things that are untrue. You are saying that Trinity will be faster in certain scenarios where it isn't.

I don't care about AMD vs Intel. I will recommend Intel or AMD depending on which is better suited for the user. However, I will not lie just just get someone to buy AMD. You, on the other hand, do this consistently even in the face of everybody trying to correct you.

 

..remember having this discussion a while back. About the Amiga. ..blablah, the graphics card chip is capable of doing this and those operations, means scrolling can be done smoother, means very efficient processor use, will be awesome once the processor develops a bit and get more ram integrated, etc, etc.

And the other guy gets angry and says.. But the intel processor has a 33Mhz processor! Are you insane! It floors the Amiga completely on everything that matters. Look at the great software-based vector graphics! I love 640kb base ram! Microsoft 3.11 rules!

That fight turned out really great for Commodore, as we know.

Seriously, though.. as people have said several times. No one is claiming that synthetic x86 performance on the amd-processors are better than intel. They're not. On synthetic tests, the latest amd processor is floored by the latest intel processors. ..even if the former draws half the power, admittedly.

But what we're claiming, with perfectly good facts to back up, is that the trinity processors will be "good enough" for office-use.. and apparently also good enough for gaming. We're practically speaking looking at about the distance from a standard clocked 2.6Ghz amd and up to a 3.3Ghz i7 being somewhere around 60 against 75 fps in Battlefield 3/1280x720, for example, on the same graphics card. Pit an a10/dedicated rig against an i5 with a good nvidia card, and we're going to see bigger differences. Take a look at the submitted 3dmark11 tests for the 7660m to get an impression of how that looks. If that difference is critical for any and all users, I don't know.

And that's... kind of surprising. That it's at all nearby in the tests.. Right?

Anyway. So what the benchmarks at least shows us is that you can make a meaningful decision between peak processing power and peak battery life. And that's great for us as users.

No need to generalize more that, imo. And go off on some categorical "intel is best all the time on everything". Since that will be incorrect.

 

Well, re-read the first couple pages. It's like saying "The HD4000 is faster than the 7660g" and then using "well, the HD4000 good enough for most people" as justification. Great, but that doesn't make the first statement true.

Battery life/price/etc. are all valid considerations. So just discuss that instead of making up false statements. (Not directed at you, nipsen)

 

out of what I've seen so far, the GPU in this chip would run at much lower clocks than its desktop equivalent, where as the HD4000 is at its maximum in both mobile and desktop variants. This tells me that overclock could be possible and would bring big improvements, where as that's it for the Intels GPU. Aside from their crappy driver support.

- for the 3D modeling usage that the OP would be doing, I cant justify going with an i5 only, seriously. Note driver support is mandatory here.

as far as the CPU performance goes, I think this chip does something like 22 Gflops or so (feel free to correct if needed), so that's like a bit over the fastest* Core 2 Duos but would beat them out on multitasking for sure. Lots of people still using C2D just fine, if you're concerned about the "average Joe type of computer use" ... I run Autocad 2007 fine on my nVidia laptop in my sig, though that video card is probably 3 times slower even with the overclock that I run on it, as it uses 64bit bus vs. its 128bit variant in 15.6" laptops.


(* - not including the heavily overclocked X9100 E0 stepping at some 4GHz or so, lol)

EDIT: Please dont get me wrong, if I were to configure a gaming laptop or a serious workstation then I'd be picking Intel CPU + dedicated GPU for sure, lol. But that's not the case here.

 

You are a nitwit, who try to sell Intel APUs claiming only Intel good for CPU tasks and all what is matters CPU power. In this topic I was showing a multithreaded situation where intel i5 have no chance against A10 and sorry if you do not like that!!! I'll keep going strongly against you and keep highlighting (as in my most of the posts) if there is difference between CPUs, that is ain't noticeable for regular users and can benefit more from faster GPU and lower power consumption and spending the saved money for SSD. This what I have been trying to explain in all of my posts, but you understanding in different way, because you do not like my truth!
And if you think, I'm short of AMD man, than I tell you No. I only had Intel computers from Nineties-2011 and I'll keep with AMD as long as they have better products for laptops. Once the situation turns back, I'll go again with Intel.

 

Where are numbers proving that the AMD APU is better? Also, lets keep from getting personal shall we?

 

First let's make clear why APUs made for!? With short answer; Integrating CPU and GPU together for less power usage which will suit better for laptops. +Saving money
So which APU better if one play games significantly better and stands with half-power use? Price point if matters for someone?

link

What Intel doing nowadays? Improving GPU performance crazy, minor CPU updates, halving power consumption. They also understand what is their lag .

 

And I agree for the most part. But OP is clearly asking for 3d rendering performance. Let's look at this thread again, shall we?

Not according to Cinebench and Autodesk.

The i5 would encode the movie faster and play the game slower due to the faster CPU and slower GPU. Saying it has "no chance" is garbage.

Remember when you said that Trinity is better at encoding movies?
Wrong, wrong and wrong. If you notice, I actually recommended Trinity over IVB in that thread because it was better suited for the OPs needs.

Yeah, seems like your "truth" isn't very close to the truth. For some reason, you can't accept that there are situations where Intel would be better suited for the user. Every time Trinity and IVB are brought up you go on an AMD marketing spiel and make stuff up even if it's irrelevant to the original post.

Oh, and battery life is actually quite similar between Trinity and Ivy Bridge.

 

I`m not shure what i5 processor you were looking at but here are some food for thoughts

 

Except OP does not care about games apparently. Power? You took a i7 with a conventional HDD and pitted it against an AMD APU that had a SSD. There itself you had a 2-3 watt power discrepancy. Those three laptops should have had ONLY the CPU as a variable. Also, where are they getting their 1080p numbers from? The ref laptop had a 768 screen. External? They haven't mentioned that either. That review reeks of mistakes.

 

Have I ever discussed with you of single, twin core performance? Can you link some of that? Because I admit Intel i5 much faster on single, twin threaded things and never doubt that. However ain't so important, since most of programs already support more than two cores. If 4 core AMD still beaten in certain situations, difference is minimal and ain't worth spending that extra few hundred $ and loose other benefits.

You are linking all kind of outsider things from many topics, just find somewhere part of your truth, while you also want keep strictly answer for the topic starter question. Two not working together!
Not mention he do not even look back what is going on here, therefore after first few post topic turned into Intel vs AMD battle as always.

 

I wonder why the op hasn't said anything else in response to their thread? Everyone has provided more than enough information to help point her/him in the right direction. Without knowing anything else, we are simply going in circles. Both Intel & AMD make fine products both with their own advantages & disadvantages. Choose the right tool for the job at the price you can afford. You can't have everything all at once all the time, compromise is unavoidable.



Sent from my SPH-M580 using Tapatalk

 

Now this has me really confused on which one to go for... AMD or intel.. ohh well back to square 1

 

Ok, A lot of great info here, and now you and I am confused, Can anyone answer that the intel will get a heavy task done 10 second sooner or 5 minutes sooner. I can wait seconds for a lower price chip. But if we are talking minutes, then it's worth paying for an i5.

Thanks, Ed

 

Define "heavy task".

 

Sorry for reviving this from the dead but after reading through the thread i'm even more confused on which processor to go for.

laptop will mainly be used for Gaming (Total war series, starcraft, few online MMO's)/Musicmaking (Fruityloops/Reason)

Rome II Recommended Specs - Total War Wiki

I play lots of oldschool shooters so i'm used to playing on lowest settings. Sometimes I play MMO and afk watch movies.

Anyone can chime in ?

 

i5 "M" CPU minimum. Don't get a "U" or "Y" (i.e. i5-4200U). And get something with a dedicated GPU. A 740m at minimum.