src/make_mamba_pptx.py

import os
import json
from pptx import Presentation
from pptx.util import Inches, Pt
from pptx.enum.text import PP_ALIGN
from pptx.dml.color import RGBColor
from pptx.enum.shapes import MSO_SHAPE

def create_presentation():
    pptx_path = "/home/adminuser/aiops_pocs/MAMBA_7B/CODESTRAL_MAMBA_7B_VS_QWEN_7B_COMPARISON.pptx"
    json_path = "/home/adminuser/aiops_pocs/MAMBA_7B/mamba_vs_qwen_results.json"
    
    with open(json_path, "r", encoding="utf-8") as f:
        benchmarks = json.load(f)

    prs = Presentation()
    prs.slide_width = Inches(13.333)
    prs.slide_height = Inches(7.5)

    # Color Palette
    COLOR_BG = RGBColor(15, 23, 42)        # Slate 900
    COLOR_CARD = RGBColor(30, 41, 59)      # Slate 800
    COLOR_ACCENT = RGBColor(6, 182, 212)    # Cyan 500
    COLOR_MAMBA = RGBColor(16, 185, 129)   # Emerald 500
    COLOR_QWEN = RGBColor(59, 130, 246)    # Blue 500
    COLOR_TEXT = RGBColor(248, 250, 252)   # Slate 50
    COLOR_MUTED = RGBColor(148, 163, 184)  # Slate 400

    def apply_bg(slide):
        bg = slide.shapes.add_shape(MSO_SHAPE.RECTANGLE, 0, 0, Inches(13.333), Inches(7.5))
        bg.fill.solid()
        bg.fill.fore_color.rgb = COLOR_BG
        bg.line.fill.background()

    def add_header(slide, title_text, category_text="BENCHMARK ANALYSIS"):
        tb = slide.shapes.add_textbox(Inches(0.8), Inches(0.4), Inches(11.7), Inches(0.9))
        tf = tb.text_frame
        tf.word_wrap = True
        
        p_cat = tf.paragraphs[0]
        p_cat.text = category_text.upper()
        p_cat.font.size = Pt(11)
        p_cat.font.bold = True
        p_cat.font.color.rgb = COLOR_ACCENT
        
        p_title = tf.add_paragraph()
        p_title.text = title_text
        p_title.font.size = Pt(24)
        p_title.font.bold = True
        p_title.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 1: Title Slide
    # -------------------------------------------------------------
    slide1 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide1)
    
    tb = slide1.shapes.add_textbox(Inches(1.0), Inches(2.2), Inches(11.3), Inches(3.2))
    tf = tb.text_frame
    tf.word_wrap = True
    
    p1 = tf.paragraphs[0]
    p1.text = "TECHNICAL BENCHMARK & ARCHITECTURAL EVALUATION"
    p1.font.size = Pt(14)
    p1.font.bold = True
    p1.font.color.rgb = COLOR_ACCENT
    
    p2 = tf.add_paragraph()
    p2.text = "Codestral Mamba 7B vs. Qwen 2.5 7B"
    p2.font.size = Pt(38)
    p2.font.bold = True
    p2.font.color.rgb = COLOR_TEXT
    
    p3 = tf.add_paragraph()
    p3.text = "Empirical comparison of Selective State Space Models (SSM) vs Multi-Head Self-Attention Transformers across 10 Technical Benchmarks on NVIDIA H200 GPUs"
    p3.font.size = Pt(16)
    p3.font.color.rgb = COLOR_MUTED
    
    # -------------------------------------------------------------
    # SLIDE 2: Executive Summary
    # -------------------------------------------------------------
    slide2 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide2)
    add_header(slide2, "Executive Summary & Core Findings", "Overview")
    
    cards = [
        ("42.2% Faster Completion", "Codestral Mamba 7B achieved an average query latency of 4.28s vs Qwen 2.5's 7.40s, delivering ultra-fast time-to-first-token.", COLOR_MAMBA),
        ("~195 Tokens / Sec Speed", "Both models saturated hardware limits at ~194.8 t/s (Mamba) vs ~193.3 t/s (Qwen) on NVIDIA H200 NVL GPUs.", COLOR_ACCENT),
        ("2.25x Explanation Density", "Qwen 2.5 7B generated 4,347 chars/response vs Mamba's 1,930 chars, providing rich docstrings & edge cases.", COLOR_QWEN),
        ("Constant O(1) Memory State", "Mamba maintains a fixed recurrent state buffer, eliminating KV-cache VRAM expansion at long context (up to 256k).", COLOR_MAMBA)
    ]
    
    for idx, (title, desc, color) in enumerate(cards):
        row = idx // 2
        col = idx % 2
        x = Inches(0.8 + col * 5.9)
        y = Inches(1.6 + row * 2.6)
        
        card = slide2.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, x, y, Inches(5.6), Inches(2.3))
        card.fill.solid()
        card.fill.fore_color.rgb = COLOR_CARD
        card.line.color.rgb = color
        card.line.width = Pt(1.5)
        
        tf = card.text_frame
        tf.word_wrap = True
        p_t = tf.paragraphs[0]
        p_t.text = title
        p_t.font.size = Pt(18)
        p_t.font.bold = True
        p_t.font.color.rgb = color
        
        p_d = tf.add_paragraph()
        p_d.text = desc
        p_d.font.size = Pt(13)
        p_d.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 3: Architectural Mechanics (SSM vs Transformer)
    # -------------------------------------------------------------
    slide3 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide3)
    add_header(slide3, "Architectural Mechanics: Selective SSM vs. Self-Attention", "Deep Dive")
    
    # Left Card: Mamba
    c1 = slide3.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(0.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c1.fill.solid()
    c1.fill.fore_color.rgb = COLOR_CARD
    c1.line.color.rgb = COLOR_MAMBA
    c1.line.width = Pt(2)
    tf1 = c1.text_frame
    tf1.word_wrap = True
    p = tf1.paragraphs[0]
    p.text = "🟢 Codestral Mamba 7B (Selective SSM S6)"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = COLOR_MAMBA
    
    bullets1 = [
        "• Core Mechanism: Input-dependent Selective State Space Model (S6).",
        "• Time Complexity: O(N) Linear scaling with sequence length.",
        "• Hardware Scanning: Parallel associative scan algorithm utilizing SRAM.",
        "• Memory Footprint: Recurrent O(1) constant-size hidden state buffer.",
        "• Key Advantage: Zero KV-cache VRAM inflation during 256k long-context inference."
    ]
    for b in bullets1:
        p = tf1.add_paragraph()
        p.text = b
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    # Right Card: Qwen Transformer
    c2 = slide3.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(6.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c2.fill.solid()
    c2.fill.fore_color.rgb = COLOR_CARD
    c2.line.color.rgb = COLOR_QWEN
    c2.line.width = Pt(2)
    tf2 = c2.text_frame
    tf2.word_wrap = True
    p = tf2.paragraphs[0]
    p.text = "🔵 Qwen 2.5 7B (Multi-Head Self-Attention)"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = COLOR_QWEN
    
    bullets2 = [
        "• Core Mechanism: Multi-Head Self-Attention with Rotary Position Embeddings (RoPE).",
        "• Time Complexity: O(N²) Quadratic attention scaling.",
        "• Hardware Optimization: FlashAttention-2 & deep Tensor Core integration.",
        "• Memory Footprint: Grows linearly per token (KV Cache memory inflation).",
        "• Key Advantage: Deep global contextual reasoning, extensive docstring synthesis."
    ]
    for b in bullets2:
        p = tf2.add_paragraph()
        p.text = b
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 4: Quantitative Performance Table
    # -------------------------------------------------------------
    slide4 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide4)
    add_header(slide4, "Quantitative Benchmark Performance Summary", "Metrics")
    
    rows = len(benchmarks) + 1
    cols = 7
    table_shape = slide4.shapes.add_table(rows, cols, Inches(0.8), Inches(1.5), Inches(11.7), Inches(5.3))
    table = table_shape.table
    
    headers = ["ID", "Technical Task Title", "Mamba Time", "Qwen Time", "Mamba Speed", "Qwen Speed", "Output Delta"]
    widths = [Inches(0.6), Inches(4.5), Inches(1.3), Inches(1.3), Inches(1.3), Inches(1.3), Inches(1.4)]
    for idx, w in enumerate(widths):
        table.columns[idx].width = w

    for c_idx, h in enumerate(headers):
        cell = table.cell(0, c_idx)
        cell.fill.solid()
        cell.fill.fore_color.rgb = COLOR_CARD
        p = cell.text_frame.paragraphs[0]
        p.text = h
        p.font.size = Pt(11)
        p.font.bold = True
        p.font.color.rgb = COLOR_ACCENT

    for r_idx, b in enumerate(benchmarks, start=1):
        m = b["mamba"]
        q = b["qwen"]
        delta = f"+{(q['char_len'] - m['char_len'])/m['char_len']*100:.0f}% Qwen"
        
        vals = [
            f"{b['id']:02d}",
            b["title"],
            f"{m['time_sec']:.2f}s",
            f"{q['time_sec']:.2f}s",
            f"{m['tps']} t/s",
            f"{q['tps']} t/s",
            delta
        ]
        for c_idx, val in enumerate(vals):
            cell = table.cell(r_idx, c_idx)
            cell.fill.solid()
            cell.fill.fore_color.rgb = RGBColor(20, 30, 48) if r_idx % 2 == 0 else COLOR_BG
            p = cell.text_frame.paragraphs[0]
            p.text = val
            p.font.size = Pt(10)
            p.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 5: Benchmark Category 1 - Algorithms & Data Structures
    # -------------------------------------------------------------
    slide5 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide5)
    add_header(slide5, "Benchmark Spotlight: Algorithms & Data Structures", "Test 01 & 02")
    
    # Test 1 Card
    c1 = slide5.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(0.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c1.fill.solid()
    c1.fill.fore_color.rgb = COLOR_CARD
    tf1 = c1.text_frame
    tf1.word_wrap = True
    p = tf1.paragraphs[0]
    p.text = "Test 01: Lock-Free LRU Cache in Python"
    p.font.size = Pt(16)
    p.font.bold = True
    p.font.color.rgb = COLOR_ACCENT
    
    p = tf1.add_paragraph()
    p.text = "• Mamba (7.16s | 1,504 chars):\nLeveraged collections.OrderedDict for a minimal, working 15-line implementation. Extremely fast & compact.\n\n• Qwen 2.5 (24.97s | 4,970 chars):\nBuilt full double-linked node class, explicit generic typing, thread safety locks, comprehensive docstrings & edge cases."
    p.font.size = Pt(12)
    p.font.color.rgb = COLOR_TEXT
    
    # Test 2 Card
    c2 = slide5.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(6.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c2.fill.solid()
    c2.fill.fore_color.rgb = COLOR_CARD
    tf2 = c2.text_frame
    tf2.word_wrap = True
    p = tf2.paragraphs[0]
    p.text = "Test 02: Red-Black Tree Balancing in C++"
    p.font.size = Pt(16)
    p.font.bold = True
    p.font.color.rgb = COLOR_ACCENT
    
    p = tf2.add_paragraph()
    p.text = "• Mamba (6.56s | 4,200 chars):\nProduced clean C++ rotation logic (left/right rotate) and color fixup helper methods directly without boilerplate.\n\n• Qwen 2.5 (8.19s | 5,793 chars):\nGenerated full C++ template struct, explicit enum Color { RED, BLACK }, driver main() function, and memory destruction logic."
    p.font.size = Pt(12)
    p.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 6: Benchmark Category 2 - System Engineering & eBPF
    # -------------------------------------------------------------
    slide6 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide6)
    add_header(slide6, "Benchmark Spotlight: Systems Engineering & eBPF", "Test 03 & 04")
    
    # Test 3 Card
    c1 = slide6.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(0.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c1.fill.solid()
    c1.fill.fore_color.rgb = COLOR_CARD
    tf1 = c1.text_frame
    tf1.word_wrap = True
    p = tf1.paragraphs[0]
    p.text = "Test 03: Asyncio WebSockets Gateway"
    p.font.size = Pt(16)
    p.font.bold = True
    p.font.color.rgb = COLOR_ACCENT
    
    p = tf1.add_paragraph()
    p.text = "• Mamba (6.13s | 3,398 chars):\nConcise Python asyncio server with heartbeat ping/pong & token check middleware.\n\n• Qwen 2.5 (5.46s | 4,874 chars):\nFull async server architecture with sliding-window rate limiting, signal handlers for graceful shutdown, and client connection registry."
    p.font.size = Pt(12)
    p.font.color.rgb = COLOR_TEXT
    
    # Test 4 Card
    c2 = slide6.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(6.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c2.fill.solid()
    c2.fill.fore_color.rgb = COLOR_CARD
    tf2 = c2.text_frame
    tf2.word_wrap = True
    p = tf2.paragraphs[0]
    p.text = "Test 04: Linux eBPF Packet Tracing (C/BCC)"
    p.font.size = Pt(16)
    p.font.bold = True
    p.font.color.rgb = COLOR_ACCENT
    
    p = tf2.add_paragraph()
    p.text = "• Mamba (3.32s | 1,147 chars):\nShort eBPF C snippet hooking kprobe/sys_enter_connect with basic BCC python print loop.\n\n• Qwen 2.5 (4.56s | 2,947 chars):\nDetailed C eBPF kernel program using BPF_HASH maps, IP byte-order conversions, error checking, and formatted BCC CLI table output."
    p.font.size = Pt(12)
    p.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 7: Strategic Comparison - Codestral Mamba 7B
    # -------------------------------------------------------------
    slide7 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide7)
    add_header(slide7, "Model Profile: Codestral Mamba 7B", "Mistral AI")
    
    c1 = slide7.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(0.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c1.fill.solid()
    c1.fill.fore_color.rgb = COLOR_CARD
    c1.line.color.rgb = COLOR_MAMBA
    tf1 = c1.text_frame
    tf1.word_wrap = True
    p = tf1.paragraphs[0]
    p.text = "🌟 Core Strengths & Advantages"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = COLOR_MAMBA
    
    b1 = [
        "1. Ultra-Low Latency: 42.2% faster average response completion.",
        "2. Constant O(1) Memory State: Eliminates KV-cache VRAM expansion at long context (up to 256k tokens).",
        "3. High Code Autocomplete Efficiency: Delivers direct, fluff-free code snippets instantly.",
        "4. High Throughput: ~194.8 t/s generation speed on H200 GPUs."
    ]
    for b in b1:
        p = tf1.add_paragraph()
        p.text = b
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    c2 = slide7.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(6.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c2.fill.solid()
    c2.fill.fore_color.rgb = COLOR_CARD
    c2.line.color.rgb = RGBColor(239, 68, 68)
    tf2 = c2.text_frame
    tf2.word_wrap = True
    p = tf2.paragraphs[0]
    p.text = "⚠️ Known Bottlenecks & Trade-offs"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = RGBColor(239, 68, 68)
    
    b2 = [
        "1. Minimal Inline Documentation: Frequently omits docstrings, type hints, and code comments.",
        "2. Concise Edge-Case Handling: May require follow-up prompts to handle complex exception paths.",
        "3. Higher Recurrent State Complexity: Requires custom SSM CUDA kernels for peak training."
    ]
    for b in b2:
        p = tf2.add_paragraph()
        p.text = b
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 8: Strategic Comparison - Qwen 2.5 7B Instruct
    # -------------------------------------------------------------
    slide8 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide8)
    add_header(slide8, "Model Profile: Qwen 2.5 7B Instruct", "Alibaba Cloud")
    
    c1 = slide8.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(0.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c1.fill.solid()
    c1.fill.fore_color.rgb = COLOR_CARD
    c1.line.color.rgb = COLOR_QWEN
    tf1 = c1.text_frame
    tf1.word_wrap = True
    p = tf1.paragraphs[0]
    p.text = "🌟 Core Strengths & Advantages"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = COLOR_QWEN
    
    b1 = [
        "1. Exhaustive Code Density: 2.25x higher output character count (4,347 chars avg).",
        "2. Complete Edge-Case Coverage: Includes explicit error checking, logging, and type hints.",
        "3. Superior Explanatory Power: Accompanies code with thorough architectural breakdowns.",
        "4. Versatile Multi-Domain Logic: Excellent performance across math proofs, code & general QA."
    ]
    for b in b1:
        p = tf1.add_paragraph()
        p.text = b
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    c2 = slide8.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(6.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c2.fill.solid()
    c2.fill.fore_color.rgb = COLOR_CARD
    c2.line.color.rgb = RGBColor(239, 68, 68)
    tf2 = c2.text_frame
    tf2.word_wrap = True
    p = tf2.paragraphs[0]
    p.text = "⚠️ Known Bottlenecks & Trade-offs"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = RGBColor(239, 68, 68)
    
    b2 = [
        "1. Higher Total Response Time: Takes 7.40s avg due to generating comprehensive explanations.",
        "2. Transformer Memory Expansion: O(N) KV-cache growth at extreme long contexts (32k+ tokens)."
    ]
    for b in b2:
        p = tf2.add_paragraph()
        p.text = b
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    # -------------------------------------------------------------
    # SLIDE 9: Production Deployment Roadmap
    # -------------------------------------------------------------
    slide9 = prs.slides.add_slide(prs.slide_layouts[6])
    apply_bg(slide9)
    add_header(slide9, "Strategic Production Deployment Roadmap", "Recommendations")
    
    c1 = slide9.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(0.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c1.fill.solid()
    c1.fill.fore_color.rgb = COLOR_CARD
    c1.line.color.rgb = COLOR_MAMBA
    tf1 = c1.text_frame
    tf1.word_wrap = True
    p = tf1.paragraphs[0]
    p.text = "🚀 Deploy Codestral Mamba 7B For:"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = COLOR_MAMBA
    
    recs1 = [
        "1. Real-Time IDE Inline Autocomplete:\nInstant sub-second code completions where speed is critical.",
        "2. Massive Repository Context Processing:\nProcessing 100k+ token codebases without VRAM KV-cache exhaustion.",
        "3. High-Throughput Streaming API Microservices:\nLow-cost, low-latency microservice integrations."
    ]
    for r in recs1:
        p = tf1.add_paragraph()
        p.text = r
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    c2 = slide9.shapes.add_shape(MSO_SHAPE.ROUNDED_RECTANGLE, Inches(6.8), Inches(1.6), Inches(5.6), Inches(5.2))
    c2.fill.solid()
    c2.fill.fore_color.rgb = COLOR_CARD
    c2.line.color.rgb = COLOR_QWEN
    tf2 = c2.text_frame
    tf2.word_wrap = True
    p = tf2.paragraphs[0]
    p.text = "🚀 Deploy Qwen 2.5 7B Instruct For:"
    p.font.size = Pt(18)
    p.font.bold = True
    p.font.color.rgb = COLOR_QWEN
    
    recs2 = [
        "1. Full-Stack System Architecture Design:\nGenerating complete production microservices with docstrings.",
        "2. Automated Refactoring & Unit Test Suites:\nCreating high-coverage pytest/unittest suites with mocks.",
        "3. Complex Debugging & Security Auditing:\nComprehensive race condition and memory leak repair."
    ]
    for r in recs2:
        p = tf2.add_paragraph()
        p.text = r
        p.font.size = Pt(13)
        p.font.color.rgb = COLOR_TEXT

    prs.save(pptx_path)
    print("Successfully generated PowerPoint presentation:", pptx_path)

if __name__ == "__main__":
    create_presentation()