Ioana Burcea

1
Initial Observations of the Simultaneous
Multithreading Pentium 4 Processor
Nathan Tuck and Dean M. Tullsen

• Ioana Burcea

2
Agenda

• SMT proposed in research
• Intel Hyper-threading
• Methodology
• Benchmarks and experiments
• Experimental Results
• Questions?

3
SMT in Research

• Up to 8 contexts 8 way SMT
• ICOUNT 2.8 fetching policy

4
Intel Hyper-threading

• SMT in real silicon Intel Pentium 4
• Single vs. multithreaded mode

5
Methodology

• Pentium 4 2.5 GHz 512 DRAM
• RedHat 7.3 Linux 2.4.28smp
• Linux treats the system as a dual-processor
• It has a separate run queue for each virtual
  processor
• Benchmarks
• SPEC CPU2000
• NAS parallel benchmarks
• SPLASH2 (modified input)

6
Speedup for Heterogeneous Workloads

• TSMT total_execution_time / number of runs
• Speedup Tseq / TSMT
• Speedup per combination Sbench_1 Sbench_2
• At least 12 total jobs
• At least 3 runs for each job

7
Static Partitioning of Resources

• SPECINT 83 on average
• SPECFP 85 on average
• eon 71
• wupwise 72
• mcf 93
• art 97
• swim 98

8
Independent Threads
9
Parallel Multithreaded Speedup
NAS
SPLASH
10
Synchronization and Communication Speed

• Reading a value protected by a lock
• 37 million times per second
• 68 cycles lock read
• Updating a value protected by a lock
• 14.6 million times per second
• 171 cycles lock update

11
Synchronization and Communication Speed (contd)

• Loop
• result independent computation
• computation that uses result flow dependence
• Independent computation
• a loop that contains
• a load
• a float multiply
• a float add

12
Synchronization and Communication Speed (contd)
13
Heterogeneous vs. Homogeneous Workloads

• Two self copies of SPEC
• Average speedup 1.11 lt 1.20
• Integer vs. integer 1.17
• Float vs. float 1.20
• Integer vs. float 1.21

14
Compiler Interaction
Baseline?
15
Questions?

• Is resource partitioning a good approach?
• IBMs Power5 implementation?
• Other implementations?