**Revision History**
-* v0.00 05may2021 first created
-* v0.01 06may2021 initial first draft
-* v0.02 08may2021 add scenarios / use-cases
-* v0.03 09may2021 add draft image for scenario
+* v0.00 05may2022 first created
+* v0.01 06may2022 initial first draft
+* v0.02 08may2022 add scenarios / use-cases
+* v0.03 09may2022 add draft image for scenario
+* v0.04 14may2022 add appendix with other research
+* v0.05 14jun2022 update images (thanks to Veera)
**Table of Contents**
Inventing a new Scalar ISA from scratch is over a decade-long task
including simulators and compilers: OpenRISC 1200 took 12 years to
-mature. Stable ISAs require Standards and Compliance Suites that
+mature. Stable Open ISAs require Standards and Compliance Suites that
take more. A Vector or Packed SIMD ISA to reach stable *general-purpose*
-auto-vectorisation compiler support has never been achieved in the
+auto-vectorization compiler support has never been achieved in the
history of computing, not with the combined resources of ARM, Intel,
AMD, MIPS, Sun Microsystems, SGI, Cray, and many more. (*Hand-crafted
assembler and direct use of intrinsics is the Industry-standard norm
Slowly, at this point, a realisation should be sinking in that, actually,
there aren't as many really truly viable Vector ISAs out there, as the
-ones that are evolving in the general direction of Vectorisation are,
+ones that are evolving in the general direction of Vectorization are,
in various completely different ways, flawed.
**Successfully identifying a limitation marks the beginning of an
Andes in Audio DSPs, WD in HDDs and SSDs. These are all
astoundingly commercially successful
multi-billion-unit mass volume markets that almost nobody
- knows anything about. Included for completeness.
+ knows anything about, outside their specialised proprietary
+ niche. Included for completeness.
In order of least controlled to most controlled, the viable
candidates for further advancement are:
The question then becomes: with all the duplication of arithmetic
operations just to make the registers scalar or vector, why not
leverage the *existing* Scalar ISA with some sort of "context"
-or prefix that augments its behaviour? Make "Scalar instruction"
+or prefix that augments its behaviour? Separate out the
+"looping" from "thing being looped on" (the elements),
+make "Scalar instruction"
synonymous with "Vector Element instruction" and through nothing
more than contextual
augmentation the Scalar ISA *becomes* the Vector ISA.
phase is greatly simplified, reducing design complexity and leaving
plenty of headroom for further expansion.
+[[!img "svp64-primer/img/power_pipelines.svg" ]]
+
Remarkably this is not a new idea. Intel's x86 `REP` instruction
-gives the base concept, but in 1994 it was Peter Hsu, the designer
+gives the base concept, and the Z80 had something similar.
+But in 1994 it was Peter Hsu, the designer
of the MIPS R8000, who first came up with the idea of Vector-augmented
prefixing of an existing Scalar ISA. Relying on a multi-issue Out-of-Order Execution Engine,
the prefix would mark which of the registers were to be treated as
Engine. The only reason that the team did not take this forward
into a commercial product
was because they could not work out how to cleanly do OoO
-multi-issue at the time.
+multi-issue at the time (leveraging Multi-Issue is the most logical
+way to exploit the Vector-Prefix concept)
In its simplest form, then, this "prefixing" idea is a matter
of:
that require significant programming effort in other ISAs.
All of these things come entirely from "Augmentation" of the Scalar operation
-being prefixed: at no time is the Scalar operation significantly
-altered.
+being prefixed: at no time is the Scalar operation's binary pattern decoded
+differently compared to when it is used as a Scalar operation.
From there, several more "Modes" can be added, including
* saturation,
Boolean Logic in a Vector context, on top of an already-powerful
Scalar Branch-Conditional/Counter instruction
+All of these festures are added as "Augmentations", to create of
+the order of 1.5 *million* instructions, none of which decode the
+32-bit scalar suffix any differently.
+
**What is missing from Power Scalar ISA that a Vector ISA needs?**
Remarkably, very little: the devil is in the details though.
sequential carry-flag chaining of these scalar instructions.
* The Condition Register Fields of the Power ISA make a great candidate
for use as Predicate Masks, particularly when combined with
- Vectorised `cmp` and Vectorised `crand`, `crxor` etc.
+ Vectorized `cmp` and Vectorized `crand`, `crxor` etc.
It is only when looking slightly deeper into the Power ISA that
certain things turn out to be missing, and this is down in part to IBM's
so Scalar ones. Examples include that transfer operations between the
Integer and Floating-point Scalar register files were dropped approximately
a decade ago after the Packed SIMD variants were considered to be
-duplicates. With it being completely inappropriate to attempt to Vectorise
+duplicates. With it being completely inappropriate to attempt to Vectorize
a Packed SIMD ISA designed 20 years ago with no Predication of any kind,
-the Scalar ISA, a much better all-round candidate for Vectorisation is
-left anaemic.
+the Scalar ISA, a much better all-round candidate for Vectorization
+(the Scalar parts of Power ISA) is left anaemic.
A particular key instruction that is missing is `MV.X` which is
illustrated as `GPR(dest) = GPR(GPR(src))`. This horrendously
expensive instruction causing a huge swathe of Register Hazards
in one single hit is almost never added to a Scalar ISA but
is almost always added to a Vector one. When `MV.X` is
-Vectorised it allows for arbitrary
+Vectorized it allows for arbitrary
remapping of elements within a Vector to positions specified
by another Vector. A typical Scalar ISA will use Memory to
achieve this task, but with Vector ISAs the Vector Register Files are
why Matrix Multiplication Schedules may not be applied to Integer
Mul-and-Accumulate, Galois Field Mul-and-Accumulate, Logical
AND-and-OR, or any other future instruction such as Complex-Number
-Multiply-and-Accumulate that a future version of the Power ISA might
+Multiply-and-Accumulate or Abs-Diff-and-Accumulate
+that a future version of the Power ISA might
support. The flexibility is not only enormous, but the compactness
-unprecedented. RADIX2 in-place DCT Triple-loop Schedules may be created in
-around 11 instructions. The only other processors well-known to have
+unprecedented. RADIX2 in-place DCT may be created in
+around 11 instructions using the Triple-loop DCT Schedule. The only other processors well-known to have
this type of compact capability are both VLIW DSPs: TI's TMS320 Series
and Qualcom's Hexagon, and both are targetted at FFTs only.
the cost of providing "Traditional" programmabilility (Virtual Memory,
SMP) is worse than counter-productive, it's often outright impossible.
+*<blockquote>
+Similar to how GPUs achieve astounding task-dedicated
+performance by giving
+ALUs 30% of total silicon area and sacrificing the ability to run
+General-Purpose programs, Aspex, Google's Tensor Processor and D-Matrix
+likewise took this route and made the same compromise.
+</blockquote>*
+
**In short, we are in "Programmer's nightmare" territory**
Having dug a proverbial hole that rivals the Grand Canyon, and
can be proposed. These are:
* [ZOLC: Zero-Overhead Loop Control](https://ieeexplore.ieee.org/abstract/document/1692906/)
+ available [no paywall](https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.301.4646&rep=rep1&type=pdf)
* [OpenCAPI and Extra-V](https://dl.acm.org/doi/abs/10.14778/3137765.3137776)
* [Snitch](https://arxiv.org/abs/2002.10143)
dynamically at runtime.
Even when deployed on as basic a CPU as a single-issue in-order RISC
-core, the performance and power-savings were astonishing: between 20
-and **80%** reduction in algorithm completion times were achieved compared
+core, the performance and power-savings were astonishing: between 27
+and **75%** reduction in algorithm completion times were achieved compared
to a more traditional branch-speculative in-order RISC CPU. MPEG
-Decode, the target algorithm specifically picked by the researcher
+Encode's timing, the target algorithm specifically picked by the researcher
due to its high complexity with 6-deep nested loops and conditional
execution that frequently jumped in and out of at least 2 loops,
came out with an astonishing 43% improvement in completion time. 43%
concept of Hybrid PE-Memory Processing were to become a JEDEC Standard,
which would increase adoption and reduce cost, a bit more thought
is required here because ARM or Intel or MIPS might not necessarily
-be happy that a Processing Element has to execute Power ISA binaries.
+be happy that a Processing Element (PE) has to execute Power ISA binaries.
At least the Power ISA is much richer, more powerful, still RISC,
and is an Open Standard, as discussed in a earlier sections.
+A reasonable compromise as a JEDEC Standard is illustrated with
+the following diagram: a 3-way Bridge PHY that allows for full
+direct interaction between DRAM ICs, PEs, and one or more main CPUs
+(* a variant of the Northbridge and/or IBM POWER10 OMI-to-DDR5 PHY concept*).
+It is also the ideal location for a "Management Core".
+If the 3-way Bridge (4-way if connectivity to other Bridge PHYs
+is also included) does not itself have PEs built-in then the ISA
+utilised on any PE or CPU is non-critical. The only concern regarding
+mixed ISAs is that the PHY should be capable of transferring all and
+any types of "Management" packets, particularly PE Virtual Memory Management
+and Register File Control (Context-switch Management given that the PEs
+are expected to be ALU-heavy and not capable of running a full SMP Operating
+System).
+
+There is also no reason why this type of arrangement should not be deployed
+in Multi-Chip-Module (aka "Chiplet") form, giving all the advantages of
+the performance boost that goes with smaller line-drivers.
+
+<img src="/openpower/sv/bridge_phy.svg" width=600 />
+
# Transparently-Distributed Vector Processing
It is very strange to the author to be describing what amounts to a
have to be "massaged" by tools that insert intrinsics into the
source code, in order to identify the Basic Blocks that the Zero-Overhead
Loops can run. Can this be merged into standard gcc and llvm
-compilers? As intrinsics: of course. Can it become part of auto-vectorisation? Probably,
+compilers? As intrinsics: of course. Can it become part of auto-vectorization? Probably,
if an infinite supply of money and engineering time is thrown at it.
Is a half-way-house solution of compiler intrinsics good enough?
Intel, ARM, MIPS, Power ISA and RISC-V have all already said "yes" on that,
to the main CPU over OpenCAPI, and the main CPU feeds back the new
TLB entries to the PE in response.
+Also in practical terms, with the PEs anticipated to be so small as to
+make running a full SMP-aware OS impractical it will not just be their TLB
+pages that need remote management but their entire register file including
+the Program Counter will need to be set up, and the ZOLC Context as
+well. With OpenCAPI packet formats being quite large a concern is that
+the context management increases latency to the point where the premise
+of this paper is invalidated. Research is needed here as to whether a
+bare-bones microkernel
+would be viable, or a Management Core closer to the PEs (on the same
+die or Multi-Chip-Module as the PEs) would allow better bandwidth and
+reduce Management Overhead on the main CPUs. However if
+the same level of power saving as Snitch (1/6th) and
+the same sort of reduction in algorithm runtime as ZOLC (20 to 80%) is not
+unreasonable to expect, this is
+definitely compelling enough to warrant in-depth investigation.
+
**Use-case: Matrix and Convolutions**
+<img src="/openpower/sv/sv_horizontal_vs_vertical.svg" />
+
+First, some important definitions, because there are two different
+Vectorization Modes in SVP64:
+
* **Horizontal-First**: (aka standard Cray Vectors) walk
through **elements** first before moving to next **instruction**
* **Vertical-First**: walk through **instructions** before
moving to next **element**. Currently managed by `svstep`,
ZOLC may be deployed to manage the stepping, in a Deterministic manner.
+Second:
+SVP64 Draft Matrix Multiply is currently set up to arrange a Schedule
+of Multiply-and-Accumulates, suitable for pipelining, that will,
+ultimately, result in a Matrix Multiply. Normal processors are forced
+to perform "loop-unrolling" in order to achieve this same Schedule.
+SIMD processors are further forced into a situation of pre-arranging rotated
+copies of data if the Matrices are not exactly on a power-of-two boundary.
+
+The current limitation of SVP64 however is (when Horizontal-First
+is deployed, at least, which is the least number of instructions)
+that both source and destination Matrices have to be in-registers,
+in full. Vertical-First may be used to perform a LD/ST within
+the loop, covered by `svstep`, but it is still not ideal. This
+is where the Snitch and EXTRA-V concepts kick in.
+
+<img src="/openpower/sv/matrix_svremap.svg" />
+
Imagine a large Matrix scenario, with several values close to zero that
could be skipped: no need to include zero-multiplications, but a
traditional CPU in no way can help: only by loading the data through
the L1-L4 Cache and Virtual Memory Barriers is it possible to
ascertain, retrospectively, that time and power had just been wasted.
-SVP64 is able to do what is termed "Vertical-First" Vectorisation,
+SVP64 is able to do what is termed "Vertical-First" Vectorization,
combined with SVREMAP Matrix Schedules. Imagine that SVREMAP has been
extended, Snitch-style, to perform a deterministic memory-array walk of
a large Matrix.
The reason in this case for the use of Vertical-First Mode is the
conditional execution of the Multiply-and-Accumulate.
-Horizontal-First Mode is the standard Cray-Style Vectorisation:
+Horizontal-First Mode is the standard Cray-Style Vectorization:
loop on all *elements* with the same instruction before moving
-on to the next instruction. Predication needs to be pre-calculated
+on to the next instruction. Horizontal-First
+Predication needs to be pre-calculated
for the entire Vector in order to exclude certain elements from
the computation. In this case, that's an expensive inconvenience
-(similar to the problems associated with Memory-to-Memory
+(remarkably similar to the problems associated with Memory-to-Memory
Vector Machines such as the CDC Star-100).
Vertical-First allows *scalar* instructions and
Draft Image (placeholder):
-<img src="/openpower/sv/zolc_svp64_extrav.jpg" width=800 />
+<img src="/openpower/sv/zolc_svp64_extrav.svg" width=800 />
The program being executed is a simple loop with a conditional
test that ignores the multiply if the input is zero.
L1 Cache needed would be miniscule compared
to the average high-end CPU.
+**Comparison of PE-CPU to GPU-CPU interaction**
+
+The informed reader will have noted the remarkable similarity between how
+a CPU communicates with a GPU to schedule tasks, and the proposed
+architecture. CPUs schedule tasks with GPUs as follows:
+
+* User-space program encounters an OpenGL function, in the
+ CPU's ISA.
+* Proprietary GPU Driver, still in the CPU's ISA, prepares a
+ Shader Binary written in the GPU's ISA.
+* GPU Driver wishes to transfer both the data and the Shader Binary
+ to the GPU. Both may only do so via Shared Memory, usually
+ DMA over PCIe (assuming a PCIe Graphics Card)
+* GPU Driver which has been running CPU userspace notifies CPU
+ kernelspace of the desire to transfer data and GPU Shader Binary
+ to the GPU. A context-switch occurs...
+
+It is almost unfair to burden the reader with further details.
+The extraordinarily convoluted procedure is as bad as it sounds. Hundreds
+of thousands of tasks per second are scheduled this way, with hundreds
+or megabytes of data per second being exchanged as well.
+
+Yet, the process is not that different from how things would work
+with the proposed microarchitecture: the differences however are key.
+
+* Both PEs and CPU run the exact same ISA. A major complexity of 3D GPU
+ and CUDA workloads (JIT compilation etc) is eliminated, and, crucially,
+ the CPU may directly execute the PE's tasks, if needed. This simply
+ is not even remotely possible on GPU Architectures.
+* Where GPU Drivers use PCIe Shared Memory, the proposed architecture
+ deploys OpenCAPI.
+* Where GPUs are a foreign architecture and a foreign ISA, the proposed
+ architecture only narrowly misses being defined as big/LITTLE Symmetric
+ Multi-Processing (SMP) by virtue of the massively-parallel PEs
+ being a bit light on L1 Cache, in favour of large ALUs and proximity
+ to Memory, and require a modest amount of "helper" assistance with
+ their Virtual Memory Management.
+* The proposed architecture has the markup points emdedded into the
+ binary programs
+ where PEs may take over from the CPU, and there is accompanying
+ (planned) hardware-level assistance at the ISA level. GPUs, which have to
+ work with a wide range of commodity CPUs, cannot in any way expect
+ ARM or Intel to add support for GPU Task Scheduling directly into
+ the ARM or x86 ISAs!
+
+On this last point it is crucial to note that SVP64 began its inspiration
+from a Hybrid CPU-GPU-VPU paradigm (like ICubeCorp's IC3128) and
+consequently has versatility that the separate specialisation of both
+GPU and CPU architectures lack.
+
**Roadmap summary of Advanced SVP64**
The future direction for SVP64, then, is:
exploiting both the Deterministic nature of ZOLC / SVREMAP
combined with the Cache-Coherent nature of OpenCAPI,
to the maximum extent possible.
+* To explore "Remote Management" of PE RADIX MMU, TLB, and
+ Context-Switching (register file transferrance) by proxy,
+ over OpenCAPI, to ensure that the distributed PEs are as
+ close to a Standard SMP model as possible, for programmers.
* To make the exploitation of this powerful solution as simple
and straightforward as possible for Software Engineers to use,
in standard common-usage compilers, gcc and llvm.
**Summary**
+There are historical and current efforts that step away from both a
+general-purpose architecture and from the practice of using compiler
+intrinsics in general-purpose compute to make programmer's lives easier.
+A classic example being the Cell Processor (Sony PS3) which required
+programmers to use DMA to schedule processing tasks. These specialist
+high-performance architectures are only tolerated for
+as long as there is no equivalent performant alternative that is
+easier to program.
+
+Combining SVP64 with ZOLC and OpenCAPI can produce an extremely powerful
+architectural base that fits well with intrinsics embedded into standard
+general-purpose compilers (gcc, llvm) as a pragmatic compromise which makes
+it useful right out the gate. Further R&D may target compiler technology
+that brings it on-par with NVIDIA, Graphcore, AMDGPU, but with intrinsics
+there is no critical product launch dependence on having such
+advanced compilers.
+
Bottom line is that there is a clear roadmap towards solving a long
standing problem facing Computer Science and doing so in a way that
reduces power consumption reduces algorithm completion time and reduces
the need for complex hardware microarchitectures in favour of much
-smaller distributed coherent Processing Elements.
+smaller distributed coherent Processing Elements with a Heterogenous ISA
+across the board.
+
+# Appendix
+
+**Samsung PIM**
+
+Samsung's
+[Processing-in-Memory](https://semiconductor.samsung.com/emea/newsroom/news/samsung-brings-in-memory-processing-power-to-wider-range-of-applications/)
+seems to be ready to launch as a
+[commercial product](https://semiconductor.samsung.com/insights/technology/pim/)
+that uses HBM as its Memory Standard,
+has "some logic suitable for AI", has parallel processing elements,
+and offers 70% reduction
+in power consumption and a 2x performance increase in speech
+recognition. Details beyond that as to its internal workings
+or programmability are minimal, however given the similarity
+to D-Matrix and Google TPU it is reasonable to place in the
+same category.
+
+* [Samsung PIM IEEE Article](https://spectrum.ieee.org/samsung-ai-memory-chips)
+ explains that there are 9 instructions, mostly FP16 arithmetic,
+ and that it is designed to "complement" AI rather than compete.
+ With only 9 instructions, 2 of which will be LOAD and STORE,
+ conditional code execution seems unlikely.
+ Silicon area in DRAM is increased by 5% for a much greater reduction
+ in power. The article notes, pointedly, that programmability will
+ be a key deciding factor. The article also notes that Samsung has
+ proposed its architecture as a JEDEC Standard.
+
+**PIM-HBM Research**
+
+[Presentation](https://ieeexplore.ieee.org/document/9073325/) by Seongguk Kim
+and associated [video](https://www.youtube.com/watch?v=e4zU6u0YIRU)
+showing 3D-stacked DRAM connected to GPUs, but notes that even HBM, due to
+large GPU size, is less advantageous than it should be. Processing-in-Memory
+is therefore logically proposed. the PE (named a Streaming Multiprocessor)
+is much more sophisticated, comprising Register File, L1 Cache, FP32, FP64
+and a Tensor Unit.
+
+<img src="/openpower/sv/2022-05-14_11-55.jpg" width=500 />
+
+**etp4hpc.eu**
+
+[ETP 4 HPC](https://etp4hpc.eu) is a European Joint Initiative for HPC,
+with an eye towards
+[Processing in Memory](https://www.etp4hpc.eu/pujades/files/ETP4HPC_WP_Processing-In-Memory_FINAL.pdf)
+
+**Salient Labs**
+
+[Research paper](https://arxiv.org/abs/2002.00281) explaining
+that they can exceed a 14 ghz clock rate Multiply-and-Accumulate
+using Photonics.
+
+**SparseLNR**
+
+[SparseLNR](https://arxiv.org/abs/2205.11622) restructures sparse
+tensor computations using loop-nest restructuring.
+
+**Additional ZOLC Resources**
+
+* <https://www.researchgate.net/publication/3351728_Zero-overhead_loop_controller_that_implements_multimedia_algorithms>