Top Banner
1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA
23
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

1

TAGE-SC-L Branch Predictors

André Seznec

INRIA/IRISA

Page 2: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

2The TAGE-SC-L branch predictorSorry, nothing really new ..

• TAGE, JILP 2006 Considered as state-of-the-art global history

predictor

• Can be augmented with small adjunct predictors

Loop predictor: CBP-2 (2006)

Statistical Corrector + Loop Predictor,

Global history CBP-3 (2011)

Local history Micro 2011

Page 3: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

3

Optimized all parameters

• Number, size, width of the tables

• Types of the histories for the statistical components

All that for decreasing the misprediction number by 3% !!

Page 4: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

4

(Main)

TAGE

Predictor

Stat.

Cor.

Prediction + Confidence

Loop Predictor

Loop Predictor

Global, local,

skeleton histories

Global, local,

skeleton histories

Page 5: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

5

TAGE: multiple tables, global history

predictor

The set of history lengths forms a geometric series

most of the storage for short history !!

{0, 2, 4, 8, 16, 32, 64, 128}

Capture correlation on very long histories

Page 6: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

6TAGE:

Tagged and prediction by the longest history matching entry

pc h[0:L1]

ctr u tag

=?

ctr u tag

=?

ctr u tag

=?

prediction

pc pc h[0:L2] pc h[0:L3]

11 1 1 1 1 1

1

1

Tagless base predictor

Page 7: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

7

=?

=?

=?1

1 1 1 1 1 1

1

1

Hit

Hit

Altpred

Pred

Miss

Page 8: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

8

Prediction computation

• General case: Longest matching component provides the prediction

• Special case: Many mispredictions on newly allocated entries: weak Ctr

On many applications, Altpred more accurate than Pred Property dynamically monitored through 4-bit counters

Page 9: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

9

A tagged table entry

• Ctr: 3-bit prediction counter

• U: 2-bit counters Was the entry recently useful ?

• Tag: partial tag

Tag CtrU

Page 10: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

10

Allocate entries on mispredictions

• Allocate entries in longer history length tables On tables with U unset

• Set Ctr to Weak and U to 0

• Limited storage budget: Allocate 2 entries for 256Kbits Allocate 1 or 2 for 32Kbits

• UNLIMITED STORAGE BUDGET: multiple entries allocated in different tables

Page 11: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

11

Managing the (U)seful counter

• Increment when avoids a misprediction (Pred = taken) & (Alt ≠ taken)

• 256K: Global decrement if « difficult » to allocate

• 32K: Probabilistic decrement when conflict

• Unlimited: don’t care

Page 12: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

12

Adjunct predictors

• TAGE tracks strong correlation with the global branch history

• Small adjunct predictors to capture some missed correlation: Loop predictor Statistical Corrector

Page 13: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

13

The loop predictor

• Predict loop with constant number of iterations: 16/32 entries less than 5 bytes per entry Capture loops with long bodies and/or

irregular internal branches

S: 1.2 % M: 1 % U:0.4%

Good tradeoff for the ChampionshipImplementation: Not that great

Page 14: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

14

The Statistical Corrector predictor

• Branches with poor correlation with global history: Sometimes better predicted by a single

wide PC indexed counter than by TAGE

• More generally, track cases such that: « In this case (PC, history, prediction),

TAGE is likely (>50 %) to mispredict »

Page 15: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

15

Small predictor: very limited budget for the SC

predictor

• Just track the statistically PC biased branches « TAGE predicts this direction on this

branch, but in most cases this was wrong »

• The corrector filter:

A small partially tagged associative table 1.5 % misp.

reduction: Much simpler than a loop predictor

Page 16: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

16

Medium predictor

« Statistically » correlated branches:•Not strongly correlated with the global history, but exhibit a bias

•better predicted by averaging than tags neural tags

« Statistically » correlated branches:•Not strongly correlated with the global history, but exhibit a bias

•better predicted by averaging than tags neural tags

Branches correlated with local history,but irregular global history pattern (on other branches)

•TAGE does not learn the pattern

Branches correlated with local history,but irregular global history pattern (on other branches)

•TAGE does not learn the pattern

Page 17: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

17MultiGehl Statistical Correlator Predictor

TAGE H

PC

Stat.Corr.

Prediction + ctr value

+

+

H + LHPCPred Gehl-like

Local hist.Local hist.

Page 18: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

18

Why does it work

• The bias table indexed with PC+TAGE output: Correct (most of the time) High counter value

Dominates, not many updates Wrong Other counters can be trained

Correlation (if it exists) can be captured

Page 19: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

19MultiGehl Statistical Correlator Predictor for the Championship

TAGE H

PC

Stat.Corr.

Prediction + ctr value

Local hist.Local hist.

+ RAS associated history+ 2 different local histories+ simple choser 6.8 % misp reduction

Page 20: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

20

« Realistic » 256 Kbits TAGE-SC-L

« Only » •12 equal size TAGE tables +•(local hist., global hist.) 4-tables SC •+ loop predictor•No history tuning

Only 2.8 % extra mispredictions

Page 21: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

21

SC for Unlimited predictor

• GEHL based SC predictor: Use any form of history information

Very long global Mutiple local « Skeleton » global history

ignore some branches Recycle old ideas from the MAC-RHSP

predictor (2004)

Page 22: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

22

SC for unlimited predictor

• 460 predictor tables + 10 choser tables Globally about 20 % less misp. than TAGE

alone

• If one removes only : The bias: 1.6 % for a single table All global history components: 3.7 % All local history components: 3.9 % The choser: 3.2 %

Page 23: 1 TAGE-SC-L Branch Predictors André Seznec INRIA/IRISA.

23

Conclusion

• TAGE-SC-L fits (nearly) all storage sizes 32Kbits ≈ 64Kbits CBP1 champion on CBP1

traces

256Kbits ≈ 512Kbits CBP3 champion on CBP4 traces

• Unlimited predictor: poTAGE-SC does better