RandomX

RandomX ("random ex") is an experimental proof of work (PoW) algorithm that uses random code execution to achieve ASIC resistance.

RandomX uses a simple low-level language (instruction set), which was designed so that any random bitstring forms a valid program.

Software implementation details and design notes are written in italics.

Virtual machine

RandomX is intended to be run efficiently and easily on a general-purpose CPU. The virtual machine (VM) which runs RandomX code attempts to simulate a generic CPU using the following set of components:

DRAM

The VM has access to 4 GiB of external memory in read-only mode. The DRAM memory blob is generated from the hash of the previous block using AES encryption (TBD). The contents of the DRAM blob change on average every 2 minutes. The DRAM blob is read with a maximum rate of 2.5 GiB/s per thread.

The DRAM blob can be generated in 0.1-0.3 seconds using 8 threads with hardware-accelerated AES and dual channel DDR3 or DDR4 memory. Dual channel DDR4 memory has enough bandwidth to support up to 16 mining threads.

MMU

The memory management unit (MMU) interfaces the CPU with the DRAM blob. The purpose of the MMU is to translate the random memory accesses generated by the random program into a DRAM-friendly access pattern, where memory reads are not bound by access latency. The MMU accepts a 32-bit address addr and outputs a 64-bit value from DRAM. The MMU splits the 4 GiB DRAM blob into 256-byte blocks. Data within one block is always read sequentially in 32 reads (32×8 bytes). Blocks are read mostly sequentially apart from occasional random jumps that happen on average every 256 blocks. The address of the next block to be read is determined 1 block ahead of time to enable efficient prefetching. The MMU uses three internal registers:

• m0 - Address of the next quadword to be read from memory (32-bit, 8-byte aligned).
• m1 - Address of the next block to be read from memory (32-bit, 256-byte aligned).
• mx - Random 32-bit counter that determines if reading continues sequentially or jumps to a random block. After each read, the read address is mixed with the counter: mx ^= addr. When the last quadword of the current block is read (the value of the m0 register ends with 0xFF), the MMU checks if the last 8 bits of mx are zero. If yes, the value of the mx register is copied into register m1.

When the value of the m1 register is changed, the memory location can be preloaded into CPU cache using the x86 PREFETCH instruction or ARM PRFM instruction. The average length of a sequential DRAM read is 64 KiB. Implicit prefetch should ensure that sequentially accessed memory is already in the cache.

Scratchpad

The VM contains a 256 KiB scratchpad, which is accessed randomly both for reading and writing. The scratchpad is split into two segments (16 KiB and 240 KiB). 75% of accesses are into the first 16 KiB.

The scratchpad access pattern mimics the usual CPU cache structure. The first 16 KiB should be covered by the L1 cache, while the remaining accesses should hit the L2 cache. In some cases, the read address can be calculated in advance (see below), which should limit the impact of L1 cache misses.

Program

The actual program is stored in a 8 KiB ring buffer structure. Each program consists of 1024 random 64-bit instructions. The ring buffer structure makes sure that the program forms a closed infinite loop.

For high-performance mining, the program should be translated directly into machine code. The whole program will typically fit into the L1 instruction cache and hot execution paths should stay in the µOP cache that is used by newer x86 CPUs. This should limit the number of front-end stalls and keep the CPU busy most of the time.

Control unit

The control unit (CU) controls the execution of the program. It reads instructions from the program buffer and sends commands to the other units. The CU contains 3 internal registers:

• pc - Address of the next instruction in the program buffer to be executed (64-bit, 8 byte aligned).
• sp - Address of the top of the stack (64-bit, 8 byte aligned).
• ic - Instruction counter contains the number of instructions to execute before terminating. The register is decremented after each instruction and the program execution stops when ic reaches 0.

Fixed number of executed instructions per program should ensure roughly equal runtime of each random program.

Stack

To simulate function calls, the VM uses a stack structure. The program interacts with the stack using the CALL, DCALL and RET instructions. The stack has unlimited size and each stack element is 64 bits wide.

Register file

The VM has 8 integer registers r0-r7 (each 64 bits wide), 8 floating point registers f0-f7 (each 64 bits wide) and 4 memory address registers g0-g3 (each 32 bits wide).

The number of registers is low enough so that they can be stored in actual hardware registers on most CPUs. The memory address registers g0-g3 can be stored in a single 128-bit vector register (xmm0-xmm15 registers for x86 and Q0-Q15 in ARM) for efficient address generation (see below).

ALU

The arithmetic logic unit (ALU) performs integer operations. The ALU can perform binary integer operations from 11 groups (ADD, SUB, MUL, DIV, AND, OR, XOR, SHL, SHR, ROL, ROR) with various operand sizes of 64, 32 or 16 bits.

FPU

The floating-point unit performs IEEE-754 compliant math using 64-bit double precision floating point numbers.

Instruction set

The 64-bit instruction is encoded as follows:

Opcode (8 bits)

There are 256 opcodes, which are distributed between various operations depending on their weight (how often they will occur in the program on average). The distribution of opcodes is following:

operation	number of opcodes
ALU operations	TBD	TBD
FPU operations	TBD	TBD
Control flow	32	12.5%

Operand a (8 bits)

a encodes the first operand, which is read from memory.

The loc(a) flag determines where the operand A is read from where the result C is saved to (see Result write-back below):

loc(a)	read A from	read address	write C to	write address
000	DRAM	32 bits	scratchpad	18 bits
001	DRAM	32 bits	scratchpad	14 bits
010	DRAM	32 bits	register x(b)	-
011	DRAM	32 bits	register x(b)	-
100	scratchpad	18 bits	scratchpad	14 bits
101	scratchpad	14 bits	scratchpad	14 bits
110	scratchpad	14 bits	register x(b)	-
111	scratchpad	14 bits	register x(b)	-

The r(a) flag encodes an integer register (r0-r7). The value of the register is first XORed with the value of the g0 register. The read address addr is then equal to the bottom 32 bits of r(a). Additionally, the value of the register and all memory address registers are rotated.

The addr value is then truncated to the required length (32, 18 or 14 bits). For reading from and writing to the scratchpad, the address is 8-byte aligned by clearing the bottom 3 bits.

If the gen flag is equal to 00, this instruction performs the Address generation step (see below).

Pseudocode:

FUNCTION GET_ADDRESS
	r(a) ^= g0
	addr = r(a)
	r(a) <<< 32
	g0 = g1
	g1 = g2
	g2 = g3
	g3 = g0
	IF gen == 0b00 THEN GENERATE_ADDRESSES
	return addr
END FUNCTION

The rotation of registers g0-g3 can be performed with a single SHUFPS x86 instruction.

Operand b (8 bits)

b encodes the second operand, which is either a register or immediate value.

loc(b)	read B from
000	register x(b)
001	register x(b)
010	register x(b)
011	register x(b)
100	register x(b)
101	register x(b)
110	imm1
111	imm1

The x(b) flag encodes a register. For ALU operations, this is an integer register (r0-r7) and for FPU operations, it's a floating point register (f0-f7).

imm1 is a 32-bit immediate value encoded within the instruction. For ALU instructions that use operands shorter than 32 bits, the value is truncated. For operands larger than 32 bits, the value is zero-extended for unsigned instructions and sign-extended for signed instructions. For FPU instructions, the value is treated as a signed 32-bit integer, first converted to a single precision floating point format and then to a double precision format.

imm0 (8 bits)

An 8-bit immediate value that is used to calculate the jump offset of the CALL and DCALL instructions.

Result writeback

All instructions take the operands A and B and produce a result C. Firstly, if C is shorter than 64 bits, it is zero-extended to 64 bits. The value of C is then written back either to the register x(b) or to the scratchpad using the same address addr from operand a (see table above).

CPUs are typically designed for a 2:1 load:store ratio, so each VM instruction performs on average 1 memory read and 0.5 write to memory.

Address generation

To ensure that the values of the memory address registers remain pseudorandom, the values of the registers are regenerated on average once in every 4 instructions.

During address generation, the 4 registers g0-g3 are combined into one 128-bit register G and the registers r(a) and x(b) are combined into a 128-bit register K. G is then encrypted with a single AES round using K as the round key.

In pseudocode:

PROCEDURE GENERATE_ADDRESSES
	G[127:96] = g3
	G[95:64] = g2
	G[63:32] = g1
	G[31:0] = g0
	K[127:64] = r(a)
	K[63:0] = x(b)
	G = AES_ROUND(G, K)
	g3 = G[127:96]
	g2 = G[95:64]
	g1 = G[63:32]
	g0 = G[31:0]
END PROCEDURE

AES_ROUND consists of the ShiftRows, SubBytes and MixColumns steps followed by XOR with K.

For x86 CPUs, address generation requires 2-3 move instructions to construct the key and a single AESENC instruction for encryption. ARM requires two separate instructions AESE and AESMC (for MixColumns). The whole address generation can run in parallel with the currently executed instruction.

ALU instructions

opcodes	instruction	signed	A width	B width	C	C width
TBD	ADD_64	no	64	64	A + B	64
TBD	ADD_32	no	32	32	A + B	32
TBD	ADD_16	no	16	16	A + B	16
TBD	SUB_64	no	64	64	A - B	64
TBD	SUB_32	no	32	32	A - B	32
TBD	SUB_16	no	16	16	A - B	16
TBD	MUL_64	no	64	64	A * B	64
TBD	MUL_32	no	32	32	A * B	64
TBD	MUL_16	no	16	16	A * B	32
TBD	IMUL_32	yes	32	32	A * B	64
TBD	IMUL_16	yes	16	16	A * B	32
TBD	DIV_64	no	64	32	A / B, A % B	64
TBD	IDIV_64	yes	64	32	A / B, A % B	64
TBD	DIV_32	no	32	16	A / B, A % B	32
TBD	IDIV_32	yes	32	16	A / B, A % B	32
TBD	AND_64	no	64	64	A & B	64
TBD	AND_32	no	32	32	A & B	32
TBD	AND_16	no	16	16	A & B	16
TBD	OR_64	no	64	64	A | B	64
TBD	OR_32	no	32	32	A | B	32
TBD	OR_16	no	16	16	A | B	16
TBD	XOR_64	no	64	64	A ^ B	64
TBD	XOR_32	no	32	32	A ^ B	32
TBD	XOR_16	no	16	16	A ^ B	16
TBD	SHL_64	no	64	6	A << B	64
TBD	SHR_64	no	64	6	A >> B	64
TBD	SAR_64	yes	64	6	A >> B	64
TBD	ROL_64	no	64	6	A <<< B	64
TBD	ROR_64	no	64	6	A >>> B	64

Division

For the division instructions, the divisor is half length of the dividend. The result C consists of both the quotient and the remainder (remainder is put the upper bits). The result of division by zero is equal to the dividend.

FPU instructions

opcodes	instruction	C
TBD	FADD	A + B
TBD	FSUB	A - B
TBD	FMUL	A * B
TBD	FDIV	A / B
TBD	FSQRT	sqrt(A)
TBD	FROUND	A

FPU instructions conform to the IEEE-754 specification, so they must give correctly rounded results. Initial rounding mode is RN (Round to Nearest). Denormal values are treated as zero.

Denormals can be disabled by setting the FTZ flag in x86 SSE and ARM Neon engines. This is done for performance reasons.

Operands loaded from memory are treated as signed 64-bit integers and converted to double precision floating point format. Operands loaded from floating point registers are used directly.

FSQRT

The sign bit of the FSQRT operand is always cleared first, so only non-negative values are used.

In x86, the SQRTSD instruction must be used. The legacy FSQRT instruction doesn't produce correctly rounded results in all cases.

FROUND

The FROUND instruction changes the rounding mode for all subsequent FPU operations depending on the two right-most bits of A:

A[1:0]	rounding mode
00	Round to Nearest (RN) mode
01	Round towards Plus Infinity (RP) mode
10	Round towards Minus Infinity (RM) mode
11	Round towards Zero (RZ) mode

The two-bit flag value exactly corresponds to bits 13-14 of the x86 MXCSR register and bits 22-33 of the ARM FPSCR register.

Control flow instructions

The following 3 control flow instructions are supported:

opcodes	instruction	function
TBD	CALL	near procedure call with a static offset
TBD	DCALL	near procedure call with a dynamic offset
TBD	RET	return from procedure

All three instructions are conditional in 75% of cases. The jump is taken only if B <= imm1. For the 25% of cases when B is equal to imm1, the jump is unconditional. In case the branch is not taken, all three instructions become "arithmetic no-op" C = A.

CALL and DCALL

Taken CALL and DCALL instructions push the values A and pc (program counter) onto the stack and then perform a forward jump relative to the value of pc. The forward offset is equal to 8 * (imm0 + 1) for the CALL instruction and 8 * ((imm0 ^ (A >> 56)) + 1) for the DCALL instruction. Maximum jump distance is therefore 256 instructions forward (this means that at least 4 correctly spaced CALL/DCALL instructions are needed to form a loop in the program).

RET

Taken RET instruction pops the return address raddr from the stack (it's the instruction following the previous CALL or DCALL), then pops a return value retval from the stack and sets C = A ^ retval. Finally, the instruction jumps back to raddr.

Program generation

The program is initialized from a 256-bit seed value using a PCG random number generator. The program is generated in this order:

1. All 1024 instructions are generated as a list of random 64-bit integers.
2. Initial values of all integer registers r0-r7 are generated as random 64-bit integers.
3. Initial values of all floating point registers f0-f7 are generated as random 64-bit signed integers converted to a double precision floating point format.
4. Initial values of all memory address registers g0-g3 are generated as random 32-bit integers.
5. The initial value of the m0 register is generated as a random 32-bit value with the last 8 bits cleared (256-byte aligned).
6. The 256 KiB cache is initialized (TBD).
7. The remaining registers are initialized as pc = 0, sp = 0, ic = 65536 (TBD), m1 = m0 + 256, mx = 0.

Result

When the program terminates (the value of ic register reaches 0), the scratchpad, the register file and the stack are hashed using the Blake2b hash function to get the final PoW value. The generation/execution can be chained multiple times to discourage mining strategies that search for programs with particular properties.