mirror of
https://github.com/ltcptgeneral/IdealRMT-DecisionTrees.git
synced 2025-09-04 14:27:23 +00:00
405 lines
12 KiB
Plaintext
405 lines
12 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 1,
|
|
"id": "58fc6db9",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import json\n",
|
|
"import math"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"id": "e07be4b3",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"f = open(\"compressed_tree.json\")\n",
|
|
"tree = json.loads(f.read())\n",
|
|
"layers = tree[\"layers\"]\n",
|
|
"classes = tree[\"classes\"]\n",
|
|
"f.close()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"id": "1516ff91",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"field_width = {\n",
|
|
"\t\"src\": 16,\n",
|
|
"\t\"dst\": 16,\n",
|
|
"\t\"protocl\": 8,\n",
|
|
"}"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "f9193827",
|
|
"metadata": {},
|
|
"source": [
|
|
"# Worst Case RMT"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"id": "5e37cfc5",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"def worst_case_rmt(tree):\n",
|
|
"\trmt = []\n",
|
|
"\tstep = 0\n",
|
|
"\n",
|
|
"\ttcam_bits = 0\n",
|
|
"\tram_bits = 0\n",
|
|
"\n",
|
|
"\tfor layer in layers:\n",
|
|
"\t\tnum_ranges = len(layers[layer])\n",
|
|
"\t\t# assume that each range requires all of 2*k prefixes when performing prefix expansion\n",
|
|
"\t\t# therefore there are 2*k * R for R ranges and width k\n",
|
|
"\t\tnum_prefixes = 2 * field_width[layer] * num_ranges\n",
|
|
"\t\tprefix_width = field_width[layer]\n",
|
|
"\n",
|
|
"\t\ttcam = {\n",
|
|
"\t\t\t\"id\": f\"{layer}_range\",\n",
|
|
"\t\t\t\"step\": step,\n",
|
|
"\t\t\t\"match\": \"ternary\",\n",
|
|
"\t\t\t\"entries\": num_prefixes,\n",
|
|
"\t\t\t\"key_size\": prefix_width\n",
|
|
"\t\t}\n",
|
|
"\t\ttcam_bits += num_prefixes * prefix_width\n",
|
|
"\n",
|
|
"\t\t# assume basic pointer reuse for metadata storage\n",
|
|
"\t\tram = {\n",
|
|
"\t\t\t\"id\": f\"{layer}_meta\",\n",
|
|
"\t\t\t\"step\": step,\n",
|
|
"\t\t\t\"match\": \"exact\",\n",
|
|
"\t\t\t\"method\": \"index\",\n",
|
|
"\t\t\t\"key_size\": math.ceil(math.log2(num_ranges)),\n",
|
|
"\t\t\t\"data_size\": len(classes)\n",
|
|
"\t\t}\n",
|
|
"\t\tram_bits += num_ranges * len(classes)\n",
|
|
"\n",
|
|
"\t\trmt.append(tcam)\n",
|
|
"\t\trmt.append(ram)\n",
|
|
"\n",
|
|
"\t\tstep += 1\n",
|
|
"\n",
|
|
"\treturn rmt, tcam_bits, ram_bits\n",
|
|
"\n",
|
|
"x, tcam_bits, ram_bits = worst_case_rmt(tree)\n",
|
|
"f = open(\"worst_case_rmt.json\", \"w+\")\n",
|
|
"f.write(json.dumps(x, indent=4))\n",
|
|
"f.close()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 5,
|
|
"id": "0dc1d6d4",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"TCAM mapping: \n",
|
|
"[1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n",
|
|
"SRAM mapping: \n",
|
|
"[1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n",
|
|
"id mapping: \n",
|
|
"[['dst_range', 'dst_meta'], ['src_range', 'src_meta'], ['protocl_range', 'protocl_meta'], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []]\n",
|
|
"TCAM bits: 13312\n",
|
|
"RAM bits: 522\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"! command python3 ideal-rmt-simulator/sim.py naive_rmt.json\n",
|
|
"print(f\"TCAM bits: {tcam_bits}\")\n",
|
|
"print(f\"RAM bits: {ram_bits}\")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "2a628655",
|
|
"metadata": {},
|
|
"source": [
|
|
"# Naive Range Expansion "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 6,
|
|
"id": "fb9febe9",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"# shamelessly stolen from: https://github.com/autolyticus/range-to-prefix/blob/master/rangetoprefix.C\n",
|
|
"\n",
|
|
"def int_to_bin(i, width):\n",
|
|
"\treturn bin(i)[2:].zfill(width)\n",
|
|
"\n",
|
|
"def increment_dc(pfx):\n",
|
|
"\tidx = pfx.find(\"*\")\n",
|
|
"\tif idx == -1:\n",
|
|
"\t\tidx = len(pfx)\n",
|
|
"\tidx = idx - 1\n",
|
|
"\t#print(pfx, pfx[:idx])\n",
|
|
"\treturn pfx[:idx] + \"*\" + pfx[idx+1:]\n",
|
|
"\t\n",
|
|
"def can_merge(pfx_a, pfx_b):\n",
|
|
"\tpfx_a = pfx_a.replace(\"*\", \"\")\n",
|
|
"\tpfx_b = pfx_b.replace(\"*\", \"\")\n",
|
|
"\treturn pfx_a[:-1] == pfx_b[:-1] and pfx_a[-1] != pfx_b[-1]\n",
|
|
"\n",
|
|
"def merge(pfx_a, prefixes):\n",
|
|
"\tpfx_a = increment_dc(pfx_a)\n",
|
|
"\tprefixes[-1] = pfx_a\n",
|
|
"\n",
|
|
"\tfor i in range(len(prefixes) - 2, -1, -1):\n",
|
|
"\t\tif can_merge(prefixes[i], prefixes[i+1]):\n",
|
|
"\t\t\tprefixes.pop()\n",
|
|
"\t\t\tpfx = increment_dc(prefixes[i])\n",
|
|
"\t\t\tprefixes[i] = pfx\n",
|
|
"\n",
|
|
"def convert_range(lower, upper, width):\n",
|
|
"\tprefixes = []\n",
|
|
"\tprefix = int_to_bin(lower, width)\n",
|
|
"\tprefixes.append(prefix)\n",
|
|
"\tnorm_upper = min(upper, 2**width-1)\n",
|
|
"\tfor i in range(lower+1, norm_upper+1):\n",
|
|
"\t\tprefix = int_to_bin(i, width)\n",
|
|
"\t\tif can_merge(prefix, prefixes[-1]):\n",
|
|
"\t\t\tmerge(prefix, prefixes)\n",
|
|
"\t\telse:\n",
|
|
"\t\t\tprefixes.append(prefix)\n",
|
|
"\treturn prefixes"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 7,
|
|
"id": "55167c28",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"def naive_rmt(tree):\n",
|
|
"\trmt = []\n",
|
|
"\tstep = 0\n",
|
|
"\n",
|
|
"\ttcam_bits = 0\n",
|
|
"\tram_bits = 0\n",
|
|
"\n",
|
|
"\tfor layer in layers:\n",
|
|
"\t\tnum_prefixes = 0\n",
|
|
"\t\tprefix_width = field_width[layer]\n",
|
|
"\t\t# for each range in the layer, convert the ranges to prefixes using naive range expansion\n",
|
|
"\t\tfor r in layers[layer]:\n",
|
|
"\t\t\tif r[\"min\"] == None:\n",
|
|
"\t\t\t\tr[\"min\"] = 0\n",
|
|
"\t\t\telif r[\"max\"] == None:\n",
|
|
"\t\t\t\tr[\"max\"] = 2 ** prefix_width\n",
|
|
"\t\t\tprefixes = convert_range(r[\"min\"], r[\"max\"], prefix_width)\n",
|
|
"\t\t\tr[\"prefixes\"] = prefixes\n",
|
|
"\t\t\tnum_prefixes += len(prefixes)\n",
|
|
"\t\t\ttcam_bits += len(prefixes) * prefix_width\n",
|
|
"\n",
|
|
"\t\ttcam = {\n",
|
|
"\t\t\t\"id\": f\"{layer}_range\",\n",
|
|
"\t\t\t\"step\": step,\n",
|
|
"\t\t\t\"match\": \"ternary\",\n",
|
|
"\t\t\t\"entries\": num_prefixes,\n",
|
|
"\t\t\t\"key_size\": prefix_width,\n",
|
|
"\t\t\t\"ranges\": layers[layer]\n",
|
|
"\t\t}\n",
|
|
"\n",
|
|
"\t\tnum_ranges = len(layers[layer])\n",
|
|
"\t\t# assume no pointer reuse for metadata storage\n",
|
|
"\t\tram = {\n",
|
|
"\t\t\t\"id\": f\"{layer}_meta\",\n",
|
|
"\t\t\t\"step\": step,\n",
|
|
"\t\t\t\"match\": \"exact\",\n",
|
|
"\t\t\t\"method\": \"index\",\n",
|
|
"\t\t\t\"key_size\": math.ceil(math.log2(num_ranges)),\n",
|
|
"\t\t\t\"data_size\": len(classes)\n",
|
|
"\t\t}\n",
|
|
"\t\tram_bits += num_ranges * len(classes)\n",
|
|
"\n",
|
|
"\t\trmt.append(tcam)\n",
|
|
"\t\trmt.append(ram)\n",
|
|
"\n",
|
|
"\t\tstep += 1\n",
|
|
"\n",
|
|
"\treturn rmt, tcam_bits, ram_bits\n",
|
|
"\n",
|
|
"x, tcam_bits, ram_bits = naive_rmt(tree)\n",
|
|
"f = open(\"naive_rmt.json\", \"w+\")\n",
|
|
"f.write(json.dumps(x, indent=4))\n",
|
|
"f.close()\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"id": "48011528",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"TCAM mapping: \n",
|
|
"[1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n",
|
|
"SRAM mapping: \n",
|
|
"[1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n",
|
|
"id mapping: \n",
|
|
"[['dst_range', 'dst_meta'], ['src_range', 'src_meta'], ['protocl_range', 'protocl_meta'], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []]\n",
|
|
"TCAM bits: 3584\n",
|
|
"RAM bits: 522\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"! command python3 ideal-rmt-simulator/sim.py naive_rmt.json\n",
|
|
"print(f\"TCAM bits: {tcam_bits}\")\n",
|
|
"print(f\"RAM bits: {ram_bits}\")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 9,
|
|
"id": "64b7271e",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"# for this technique, we note that given disjoint ranges [0,a][a,b],[b,c] ...\n",
|
|
"# then if using a TCAM that selects the first matching prefix, then [0,a],[0,b],[0,c] would be equivalent\n",
|
|
"# this is because if for some k<a, even though the range [0,b] could be selected, as long as the prefixes for [0,a] are before [0,b] then the correct prefix will still be selected\n",
|
|
"\n",
|
|
"def priority_aware(tree):\n",
|
|
"\trmt = []\n",
|
|
"\tstep = 0\n",
|
|
"\n",
|
|
"\ttcam_bits = 0\n",
|
|
"\tram_bits = 0\n",
|
|
"\n",
|
|
"\tfor layer in layers:\n",
|
|
"\t\tnum_prefixes = 0\n",
|
|
"\t\tprefix_width = field_width[layer]\n",
|
|
"\t\t# for each range, run the regular prefix expansion, and also the prefix expansion setting the minimum to 0\n",
|
|
"\t\t# then check which set of prefixes would be better\n",
|
|
"\t\t# we will assume the ranges are already disjoin and in the correct order\n",
|
|
"\t\tfor r in layers[layer]:\n",
|
|
"\t\t\tif r[\"min\"] == None:\n",
|
|
"\t\t\t\tr[\"min\"] = 0\n",
|
|
"\t\t\telif r[\"max\"] == None:\n",
|
|
"\t\t\t\tr[\"max\"] = 2 ** prefix_width\n",
|
|
"\t\t\tregular_prefixes = convert_range(r[\"min\"], r[\"max\"], prefix_width)\n",
|
|
"\t\t\tzero_start_prefixes = convert_range(0, r[\"max\"], prefix_width)\n",
|
|
"\n",
|
|
"\t\t\tif len(regular_prefixes) <= len(zero_start_prefixes):\n",
|
|
"\t\t\t\tpfx_type = \"exact\"\n",
|
|
"\t\t\t\tprefixes = regular_prefixes\n",
|
|
"\t\t\telse:\n",
|
|
"\t\t\t\tpfx_type = \"zero\"\n",
|
|
"\t\t\t\tprefixes = zero_start_prefixes\n",
|
|
"\n",
|
|
"\t\t\tr[\"prefixes\"] = prefixes\n",
|
|
"\t\t\tr[\"prefix_type\"] = pfx_type\n",
|
|
"\t\t\tnum_prefixes += len(prefixes)\n",
|
|
"\t\t\ttcam_bits += len(prefixes) * prefix_width\n",
|
|
"\n",
|
|
"\t\ttcam = {\n",
|
|
"\t\t\t\"id\": f\"{layer}_range\",\n",
|
|
"\t\t\t\"step\": step,\n",
|
|
"\t\t\t\"match\": \"ternary\",\n",
|
|
"\t\t\t\"entries\": num_prefixes,\n",
|
|
"\t\t\t\"key_size\": prefix_width,\n",
|
|
"\t\t\t\"ranges\": layers[layer]\n",
|
|
"\t\t}\n",
|
|
"\n",
|
|
"\t\tnum_ranges = len(layers[layer])\n",
|
|
"\t\t# assume no pointer reuse for metadata storage\n",
|
|
"\t\tram = {\n",
|
|
"\t\t\t\"id\": f\"{layer}_meta\",\n",
|
|
"\t\t\t\"step\": step,\n",
|
|
"\t\t\t\"match\": \"exact\",\n",
|
|
"\t\t\t\"method\": \"index\",\n",
|
|
"\t\t\t\"key_size\": math.ceil(math.log2(num_ranges)),\n",
|
|
"\t\t\t\"data_size\": len(classes)\n",
|
|
"\t\t}\n",
|
|
"\t\tram_bits += num_ranges * len(classes)\n",
|
|
"\n",
|
|
"\t\trmt.append(tcam)\n",
|
|
"\t\trmt.append(ram)\n",
|
|
"\n",
|
|
"\t\tstep += 1\n",
|
|
"\n",
|
|
"\treturn rmt, tcam_bits, ram_bits\n",
|
|
"\n",
|
|
"x, tcam_bits, ram_bits = priority_aware(tree)\n",
|
|
"f = open(\"priority_aware.json\", \"w+\")\n",
|
|
"f.write(json.dumps(x, indent=4))\n",
|
|
"f.close()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 10,
|
|
"id": "cd706e41",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"TCAM mapping: \n",
|
|
"[1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n",
|
|
"SRAM mapping: \n",
|
|
"[1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n",
|
|
"id mapping: \n",
|
|
"[['dst_range', 'dst_meta'], ['src_range', 'src_meta'], ['protocl_range', 'protocl_meta'], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []]\n",
|
|
"TCAM bits: 2120\n",
|
|
"RAM bits: 522\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"! command python3 ideal-rmt-simulator/sim.py priority_aware.json\n",
|
|
"print(f\"TCAM bits: {tcam_bits}\")\n",
|
|
"print(f\"RAM bits: {ram_bits}\")"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": "switch",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.12.7"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|