Does such a thing as a deterministic (as in same result every run) architecture emulator exist? It is to benchmark test compilers/interpreters.
I do not mean an emulator that simply runs your program on whatever simulated architecture, but something that would compute an efficiency/speed index based on the analysis of the generated code (such as, the thing would have a deterministic value for the time taken by each instruction).
I can compute benchmark statistics on a real machine, but a deterministic result would eliminate the particularities of my machine and allow me to see the effect of small optimizations.
Intel's IACA is a static analysis tool. What is IACA and how do I use it?. But it only works for a single loop and doesn't model cache effects, only the pipeline. (And it assumes nearly-ideal OoO scheduling, I think, so probably doesn't find ROB-size limits, only front-end vs. execution port vs. loop-carried dependency latency bottlenecks). Plus IACA has some bugs in its cost model (e.g. its unlamination rules for micro-fusion of indexed addressing modes are wrong for Haswell).
AFAIK, there are no cycle accurate x86 simulators publicly available for any modern micro-architecture. We only have emulators that don't even try to run at the same speed as any real hardware, just as fast as possible, like BOCHS and qemu. I'm sure Intel and AMD have simulator software internally to validate CPU designs and model their performance, though.
You could probably assign a cycle cost to every instruction in an interpreting emulator like BOCHS and get a deterministic number, and maybe model the cache, too (there are cache simulators). It would be the same every time you ran it, but it wouldn't correspond to the running time on any real hardware!
Being deterministic is nowhere near sufficient to be interesting for tuning software. Modern x86 CPUs have a lot of microarchitectural state for out-of-order execution. We can often predict very close to how they'll run a loop (http://agner.org/optimize/, and other performance links in the x86 tag wiki), but on a larger scale there are many things that are only known by the vendors so so we couldn't write a truly accurate simulator even if we had the time. Things like branch-prediction are known in general terms, but the details have not been reverse-engineered in full detail. But branch prediction is a critical part of making a heavily pipelined CPU sustain anywhere near 3 to 4 fused-domain (front-end) uops per clock in real code.
Things get even more complicated if you want to model a multi-core machine, and SMT / HT adds lots of complexity between threads sharing a core. It's barely deterministic in the real hardware because small timing variations can lead to different threads getting farther out of sync.
To be really useful, you'd want to be able to test your code on Sandybridge, Haswell, Skylake, Bulldozer, Ryzen, and maybe Silvermont. And maybe different variants of those with different amounts of cache, and server vs. desktop where L3 / memory latency differs. (Many-core servers have significantly worse uncore latency, and lower single-threaded bandwidth even though the aggregate bandwidth is higher.)
So the whole idea of a deterministic simulator for "the x86 architecture" is weird. You could make one as simply as by giving each instruction a cost of 1 cycle, but that would be totally unrealistic.
Related
I have an x86-64 Linux program which I am attempting to optimize via perf. The perf report shows the hottest instructions are scalar conversions from double to long with a memory argument, for example:
cvttsd2si (%rax, %rdi, 8), %rcx
which corresponds to C code like:
extern double *array;
long val = (long)array[idx];
(This is an unusual bottleneck but the code itself is very unusual.)
To inform optimizations I want to know if these instructions are hot because of the load from memory, or because of the arithmetic conversion itself. What's the best way to answer this question? What other data should I collect and how should I proceed to optimize this?
Some things I have looked at already. CPU counter results show 1.5% cache misses per instruction:
30686845493287 cache-references
2140314044614 cache-misses # 6.975 % of all cache refs
52970546 page-faults
1244774326560850 instructions
194784478522820 branches
2573746947618 branch-misses # 1.32% of all branches
52970546 faults
Top-down performance monitors show we are primarily backend-bound:
frontend bound retiring bad speculation backend bound
10.1% 25.9% 4.5% 59.5%
Ad-hoc measurement with top shows all CPUs pegged at 100% suggesting we are not waiting on memory.
A final note of interest: when run on AWS EC2, the code is dramatically slower (44%) on AMD vs Intel with the same core count. (Tested on Ice Lake 8375C vs EPYC 7R13). What could explain this discrepancy?
Thank you for any help.
To inform optimizations I want to know if these instructions are hot because of the load from memory, or because of the arithmetic conversion itself. What's the best way to answer this question?
I think there is two main reason for this instruction to be slow. 1. There is a dependency chain and the latency of this instruction is a problem since the processor is waiting on it to execute other instructions. 2. There is a cache miss (saturating the memory with such instruction is improbable unless many cores are doing memory-based operations).
First of all, tracking what is going on on a specific instruction is hard (especially if the instruction is not executed a lot of time). You need to use precise events to track the root of the problem, that is, events for which the exact instruction addresses that caused the event are available. Only a (small) subset of all events are precise one.
Regarding (1), the latency of the instruction should be about 12 cycles on both architecture (although it might be slightly more on the AMD processor, I do not expect a 44% difference). The target processor are able to execute multiple instruction at the same time in a given cycle. Instructions are executed on different port and are also pipelined. The port usage matters to understand what is going on. This means all the instruction in the hot loop matters. You cannot isolate this specific instruction. Modern processors are insanely complex so a basic analysis can be tricky. On Ice Lake processors, you can measure the average port usage with events like UOPS_DISPATCHED.PORT_XXX where XXX can be 0, 1, 2_3, 4_9, 5, 6, 7_8. Only the first three matters for this instruction. The EXE_ACTIVITY.XXX events may also be useful. You should check if a port is saturated and which one. AFAIK, none of these events are precise so you can only analyse a block of code (typically the hot loop). On Zen 3, the ports are FP23 and FP45. IDK what are the useful events on this architecture (I am not very familiar with it).
On Ice Lake, you can check the FRONTEND_RETIRED.LATENCY_GE_XXX events where XXX is a power of two integer (which should be precise one so you can see if this instruction is the one impacting the events). This help you to see whether the front-end or the back-end is the limiting factor.
Regarding (2), you can check the latency of the memory accesses as well as the number of L1/L2/L3 cache hits/misses. On Ice Lake, you can use events like MEM_LOAD_RETIRED.XXX where XXX can be for example L1_MISS L1_HIT, L2_MISS, L2_HIT, L3_MISS and L3_HIT. Still on Ice Lake, t may be useful to track the latency of the memory operation with MEM_TRANS_RETIRED.LOAD_LATENCY_GT_XXX where XXX is again a power of two integer.
You can also use LLVM-MCA to simulate the scheduling of the loop instruction statically on the target architecture (do not consider branches). This is very useful to understand deeply what the scheduler can do pretty easily.
What could explain this discrepancy?
The latency and reciprocal throughput should be about the same on the two platform or at least close. That being said, for the same core count, the two certainly do not operate at the same frequency. If this is not coming from that, then I doubt this instruction is actually the problem alone (tricky scheduling issues, wrong/inaccurate profiling results, etc.).
CPU counter results show 1.5% cache misses per instruction
The thing is the cache-misses event is certainly not very informative here. Indeed, it references the last-level cache (L3) misses. Thus, it does not give any information about the L1/L2 misses (previous events do).
how should I proceed to optimize this?
If the code is latency bound, the solution is to first break any dependency chain in this loop. Unrolling the loop dans rewriting it so to make it more SIMD-friendly can help a lot to improve performance (the reciprocal throughput of this instruction is about 1 cycle as opposed to 12 for the latency so there is a room for improvements in this case).
If the code is memory bound, they you should care about data locality. Data should fit in the L1 cache if possible. There are many tips to do so but it is hard to guide you without more context. This includes for example sorting data, reordering loop iterations, using smaller data types.
There are many possible source of weird unusual unexpected behaviours that can occurs. If such a thing happens, then it is nearly impossible to understand what is going on without the exact code executed. All details matter in this case.
My question primarily applies to firestorm/icestorm (because that's the hardware I have), but I am curious about what other representative arm cores do too. Arm has strange pre- and post-incremented addressing modes. If I have (for instance) two post-incremented loads from the same register, will the second depend on the first, or is the CPU smart enough to perform them in parallel?
AFAIK the exact behaviour of the M1 execution units is mainly undocumented. Still, there is certainly a dependency chain in this case. In fact, it would be very hard to break it and the design of modern processors make this even harder: the decoders, execution units, schedulers are distinct units and it would be insane to dynamically adapt the scheduling based on the instructions executed in parallel by execution units so to be able to break the chain in this particular case. Not to mention that instructions are pipelined and it generally takes few cycles for them to be committed. Furthermore, the time of the instructions is variable based on the fetched memory location. Finally, even this would be the case, the Firestorm documents does not mention such a feedback loop (see below for the links). Another possible solution for a processor to optimize such a pattern is to fuse the microinstructions so to combine the increment and add more parallelism but this is pretty complex to do for a relatively small improvement and there is no evidence showing Firestorm can do that so far (see here for more information about Firestorm instruction fusion/elimitation).
The M1 big cores (Apple's Firestorm) are designed to be massively parallel. They have 6 ALUs per core so they can execute a lot instructions in parallel on each core (possibly at the expense of a higher latency). However, this design tends to require a lot more transistors than current mainstream x86 Intel/AMD alternative (Alderlake/XX-Cove architecture put aside). Thus, the cores operate at a significantly lower frequency so to keep the energy consumption low. This means dependency chains are significantly more expensive on such an architecture compared to others unless there are enough independent instructions to be execute in parallel on the critical path. For more information about how CPUs works please thread Modern Microprocessors - A 90-Minute Guide!. For more information about the M1 processors and especially the Firestorm architecture, please read this deep analysis.
Note that Icestorm cores are designed to be energy efficient so they are far less parallel and thus having a dependency chain should be less critical on such a core. Still, having less dependency is often a good idea.
As for other ARM processors, recent core architecture are not as parallel as Firestorm. For example, the Cortex-A77 and Neoverse V1 have "only" 4 ALUs (which is already quite good). One need to also care about the latency of each instruction actually used in a given code. This information is available on the ARM website and AFAIK not yet published for Apple processors (one need to benchmark the instructions).
As for the pre VS post increment, I expect them to take the same time (same latency and throughput), especially on big cores like Firestorm (that try to reduce the latency of most frequent instruction at the expense of more transistors). However, the actual scheduling of the instruction for a given code can cause one to be slower than the other if the latency is not hidden by other instructions.
I received an answer to this on IRC: such usage will be fairly fast (makes sense when you consider it corresponds to typical looping patterns; good if the loop-carried dependency doesn't hurt too much), but it is still better to avoid it if possible, as it takes up rename bandwidth.
Judging by the latest news, new Apple processor A11 Bionic gains more points than the mobile Intel Core i7 in the Geekbench benchmark.
As I understand, there are a lot of different tests in this benchmark. These tests simulate a different load, including the load, which can occur in everyday use.
Some people state that these results can not be compared to x86 results. They say that x86 is able to perform "more complex tasks". As an example, they lead Photoshop, video conversion, scientific calculations. I agree that the software for the ARM is often only a "lighweight" version of software for desktops. But it seems to me that this limitation is caused by the format of mobile operating systems (do your work on the go, no mouse, etc), and not by the performance of ARM.
As an opposite example, let's look at Safari. A browser is a complex program. And on the iPad Safari works just as well as on the Mac. Moreover, if we take the results of Sunspider (JS benchmark), it turns out that Safari on the iPad is gaining more points.
I think that in everyday tasks (Web, Office, Music/Films) ARM (A10X, A11) and x86 (dual core mobile Intel i7) performance are comparable and equal.
Are there any kinds of tasks where ARM really lags far behind x86? If so, what is the reason for this? What's stopping Apple from releasing a laptop on ARM? They already do same thing with migration from POWER to x86. This is technical restrictions, or just marketing?
(Intended this as a comment since this question is off topic, but it got long..).
Of course you can compare, you just need to be very careful, which most people aren't. The fact that companies publishing (or "leaking") results are biased also doesn't help much.
The common misconception is that you can compare a benchmark across two systems and get a single score for each. That ignores the fact that different systems have different optimization points, most often with regards to power (or "TDP"). What you need to look at is the power/performance curve - this graph shows how the system reacts to more power (raising the frequency, enabling more performance features, etc), and how much it contributes to its performance.
One system can win over the low power range, but lose when the available power increases since it doesn't scale that well (or even stops scaling at some point). This is usually the case with Arm, as most of these CPUs are tuned for low power, while x86 covers a larger domain and scales much better.
If you are forced to observe a single point along the graph (which is a legitimate scenario, for example if you're looking for a CPU for a low-power device), at least make sure the comparison is fair and uses the same power envelope.
There are of course other factors that must be aligned (and sometimes aren't due to negligence or an intention to cheat) - the workload should be the same (i've seen different versions compared..), the compiler should be as close as possible (although generating arm vs x86 code is already a difference, but the compiler intermediate optimizations should be similar. When comparing 2 x86 like intel and AMD you should prefer the same binary, unless you also want to allow machine specific optimizations).
Finally, the system should also be similar, which is not the case when comparing a smartphone against a pc/macbook. The memory could differ, the core count, etc. This could be legitimate difference, but it's not really related to one architecture being better than the other.
the topic is bogus, from the ISA to an application or source code there are many abstraction level and the only metric that we have (execution time, or throughput) depends on many factors that could advantage one or the other: the algorithm choices, the optimization written in source code, the compiler/interpreter implementation/optimizations, the operating system behaviour. So they are not exactly/mathematically comparable.
However, looking at the numbers, and the utility of the mobile application written by talking as a management engeneer, ARM chip seems to be capable of run quite good.
I think the only reason is inertia of standard spread around (if you note microsoft propose a variant of windows running on ARM processors, debian ARM variant are ready https://www.debian.org/distrib/netinst).
the ARMv8 cores seems close to x86/64 ones by looking at raw numbers
note i7-3770k results: https://en.wikipedia.org/wiki/Instructions_per_second#MIPS
summary of last Armv8 CPU characteristics, note the quantity of decode, dispatch, caches, and compare the last column on cortex A73 to the i7 3770k
https://en.wikipedia.org/wiki/Comparison_of_ARMv8-A_cores
intel ivy bridge characteristics:
https://en.wikichip.org/wiki/intel/microarchitectures/ivy_bridge_(client)
A75 details. https://www.anandtech.com/show/11441/dynamiq-and-arms-new-cpus-cortex-a75-a55
the topic of power consumption is complex again, the basic rule that go under all the frequency/tension rule (used and abused) over www is: transistors raise time. https://en.wikipedia.org/wiki/Rise_time
There is a fixed time delay in the switching of a transistor, this determinates the maximum frequency that a transistor could switch, and with more of them linked in a cascade way this time sums up in a nonlinear way (need some integration to demonstrate it), as a result 10 years ago to increase the GHz companies try to split in more stage the execution of an operation and runs them (operations) in a pipeline way, even inside the logical pipeline stage. https://en.wikipedia.org/wiki/Instruction_pipelining
the raise time depends of physical characteristics (materials and shape of transistors). It can be reduced by increasing the voltage, so the transistor switch faster, as the switching is associated (let me the term) to a the charge/discharge of a capacitor that trigger the transistor channel opening/closing.
These ARM chips are designed to low power applications, by changing the design they could easily gain MHz, but they will use much power, how much? again not comparable if you don't work inside a foundry and have the numbers.
an example of server applications of ARM processors that could be closer to desktop/workstation CPU as power consumption are Cavium or qualcomm Falkor CPUs, and some benchmark report that they are not bad.
I am tunning my GEMM code and comparing with Eigen and MKL. I have a system with four physical cores. Until now I have used the default number of threads from OpenMP (eight on my system). I assumed this would be at least as good as four threads. However, I discovered today that if I run Eigen and my own GEMM code on a large dense matrix (1000x1000) I get better performance using four threads instead of eight. The efficiency jumped from 45% to 65%. I think this can be also seen in this plot
https://plafrim.bordeaux.inria.fr/doku.php?id=people:guenneba
The difference is quite substantial. However, the performance is much less stable. The performance jumps around quit a bit each iteration both with Eigen and my own GEMM code. I'm surprised that Hyperthreading makes the performance so much worse. I guess this is not not a question. It's an unexpected observation which I'm hoping to find feedback on.
I see that not using hyper threading is also suggested here.
How to speed up Eigen library's matrix product?
I do have a question regarding measuring max performance. What I do now is run CPUz and look at the frequency as I'm running my GEMM code and then use that number in my code (4.3 GHz on one overclocked system I use). Can I trust this number for all threads? How do I know the frequency per thread to determine the maximum? How to I properly account for turbo boost?
The purpose of hyperthreading is to improve CPU usage for code exhibiting high latency. Hyperthreading masks this latency by treating two threads at once thus having more instruction level parallelism.
However, a well written matrix product kernel exhibits an excellent instruction level parallelism and thus exploits nearly 100% of the CPU ressources. Therefore there is no room for a second "hyper" thread, and the overhead of its management can only decrease the overall performance.
Unless I've missed something, always possible, your CPU has one clock shared by all its components so if you measure it's rate at 4.3GHz (or whatever) then that's the rate of all the components for which it makes sense to figure out a rate. Imagine the chaos if this were not so, some cores running at one rate, others at another rate; the shared components (eg memory access) would become unmanageable.
As to hyperthreading actually worsening the performance of your matrix multiplication, I'm not surprised. After all, hyperthreading is a poor-person's parallelisation technique, duplicating instruction pipelines but not functional units. Once you've got your code screaming along pushing your n*10^6 contiguous memory locations through the FPUs a context switch in response to a pipeline stall isn't going to help much. At best the other pipeline will scream along for a while before another context switch robs you of useful clock cycles, at worst all the careful arrangement of data in the memory hierarchy will be horribly mangled at each switch.
Hyperthreading is designed not for parallel numeric computational speed but for improving the performance of a much more general workload; we use general-purpose CPUs in high-performance computing not because we want hyperthreading but because all the specialist parallel numeric CPUs have gone the way of all flesh.
As a provider of multithreaded concurrency services, I have explored how hyperthreading affects performance under a variety of conditions. I have found that with software that limits its own high-utilization threads to no more that the actual physical processors available, the presence or absence of HT makes very little difference. Software that attempts to use more threads than that for heavy computational work, is likely unaware that it is doing so, relying on merely the total processor count (which doubles under HT), and predictably runs more slowly. Perhaps the largest benefit that enabling HT may provide, is that you can max out all physical processors, without bringing the rest of the system to a crawl. Without HT, software often has to leave one CPU free to keep the host system running normally. Hyperthreads are just more switchable threads, they are not additional processors.
Approximately, how many physical instructions of MIPS does an abstract algorithm operation amortize to? As for an abstract algorithm operation, I means a basic operation, such as add, divide, etc.
I see this is not a strict measuring technique :-)
Kejia
There is a list of the basic MIPS instructions here. Most of the "basic operations" that you mentioned are a single MIPS instruction or perhaps two, which probably holds true on most current CPU families.
However this does not take into account at all the architecture and performance characteristics of any of the modern CPUs. Different instructions often have diffrent completion times. Current CPUs usually implement branch prediction, instruction pipelines, memory caching, parallelisation and a whole list of other techniques to make the code execution faster.
Therefore just having the assembly code implementation of an algorithm says nothing about its execution speed. You would have to measure and profile the code on the actual hardware to obtain comparable results. In fact, some algorithms may be far more effective on certain CPUs, even within the same CPU family.
A common and rather understandable example is the effect of the instruction cache. Unrolling a loop will eliminate a number of branch operations, which intuitively makes code faster. If you run that code on a CPU of the same family with very little instruction cache memory, though, the added accesses to the main memory can make it far slower than the simple branch-based loop.
Computers are complicated. If you want to get down to this level you need to start considering what kind of CPU you are using, how well your compiler can use this CPU's instruction set, what variables are being kept in what registers, what are their bit-level representations, etc. Even then, the number of instructions not always easily maps to the actual running time. Different instructions can take different ammounts of clock cycles to execute and this is not even thinking about OS threading and your program's cache miss rate.
In the end, there is a good reason we use big-O notatoin in the first place :)
BTW, most simple operations (add, subtract) on integers should map to a single machine instruction, in case you are worried.
It depends on the CPU architecture. Some processors requires several cycles for a single instruction such as divivide, while others manage to execute all machine code instructions in a single cycle each.
It is sometimes relevant to measure an algorithm in how many floating point operations it requires. However this does not take I/O (such as reading memory) into consideration.
The speed of a CPU is sometimes provided in FLOPS (Floating Point OPerations per Second) which could help to give you a time estimate. Again, not taking I/O into consideration - and not multi-threading issues (also a very important measuring factor).
Donald Knuth addressed this very problem in writing Volume 1 of "The Art of Computer Programming".
In the preface he gives a lengthy justification for presenting algorithms in the assembly code for an imaginary machine -
... To avoid this dilemma, I have
attempted to design an "ideal"
computer called "MIX," with very
simple rules of operation ...
That way, one can talk sensibly about how many "cycles" an algorithm would take, without having to care about differences between machines, caching, latency, pipelines, or any of the other ways computers have been optimized to save time, at the expense of knowing how long they will take.