Dr. Neil Gunther
@drqz.bsky.social
Performance analyst, 'rocket scientist' & Xerox PARC alum 🇦🇺🇺🇸 Consultant, author, teacher, physics, math, stats, chem and some 🎸 performance too.
Ordinary AI output requires extraordinary human verification. (Apologies to C. Sagan)
October 23, 2025 at 11:01 PM
Ordinary AI output requires extraordinary human verification. (Apologies to C. Sagan)
Productivity: If you want to be more productive, go to sleep.
www.perfdynamics.com/Manifesto/gc...
www.perfdynamics.com/Manifesto/gc...
October 13, 2025 at 6:32 AM
Productivity: If you want to be more productive, go to sleep.
www.perfdynamics.com/Manifesto/gc...
www.perfdynamics.com/Manifesto/gc...
Out of The Queue Comes ... Amdahl’s Law!
www.perfdynamics.com/Papers/amdsy...
www.perfdynamics.com/Papers/amdsy...
September 18, 2025 at 5:10 PM
Out of The Queue Comes ... Amdahl’s Law!
www.perfdynamics.com/Papers/amdsy...
www.perfdynamics.com/Papers/amdsy...
RIP James Lovell, commander of Apollo 13 (played by Tom Hanks in the movie of the same name), has died at age 97.
August 8, 2025 at 8:57 PM
RIP James Lovell, commander of Apollo 13 (played by Tom Hanks in the movie of the same name), has died at age 97.
A. Displaying metrics, errors, etc., is not the same thing as displaying metric errors.
1. How it is (old school top)
CPU usage: 5.79 %usr, 4.51 %sys, 89.68 %idl
2. How it could be (nay, SHOULD be)
CPU usage: 5.79 ± 0.29 %usr, 4.51 ± 0.23 %sys, 89.68 ± 4.48 %idl
1. How it is (old school top)
CPU usage: 5.79 %usr, 4.51 %sys, 89.68 %idl
2. How it could be (nay, SHOULD be)
CPU usage: 5.79 ± 0.29 %usr, 4.51 ± 0.23 %sys, 89.68 ± 4.48 %idl
August 7, 2025 at 5:43 PM
A. Displaying metrics, errors, etc., is not the same thing as displaying metric errors.
1. How it is (old school top)
CPU usage: 5.79 %usr, 4.51 %sys, 89.68 %idl
2. How it could be (nay, SHOULD be)
CPU usage: 5.79 ± 0.29 %usr, 4.51 ± 0.23 %sys, 89.68 ± 4.48 %idl
1. How it is (old school top)
CPU usage: 5.79 %usr, 4.51 %sys, 89.68 %idl
2. How it could be (nay, SHOULD be)
CPU usage: 5.79 ± 0.29 %usr, 4.51 ± 0.23 %sys, 89.68 ± 4.48 %idl
Online classroom buzz www.perfdynamics.com/Classes/comm...
What Guerrilla Grads are saying about why you should become an alumnus.
Enroll here www.perfdynamics.com/Classes/sche... 🎓
#performance #engineering #capacity #management #marketplace #trends #GCAP #GDAT #PDQW
What Guerrilla Grads are saying about why you should become an alumnus.
Enroll here www.perfdynamics.com/Classes/sche... 🎓
#performance #engineering #capacity #management #marketplace #trends #GCAP #GDAT #PDQW
August 6, 2025 at 1:07 PM
Online classroom buzz www.perfdynamics.com/Classes/comm...
What Guerrilla Grads are saying about why you should become an alumnus.
Enroll here www.perfdynamics.com/Classes/sche... 🎓
#performance #engineering #capacity #management #marketplace #trends #GCAP #GDAT #PDQW
What Guerrilla Grads are saying about why you should become an alumnus.
Enroll here www.perfdynamics.com/Classes/sche... 🎓
#performance #engineering #capacity #management #marketplace #trends #GCAP #GDAT #PDQW
PSA: New open access Journal of Systems Scalability launching.
Editor-in-Chief: Carlos Juiz
www.icck.org/jss
#systems #architecture #performance #scalability
Editor-in-Chief: Carlos Juiz
www.icck.org/jss
#systems #architecture #performance #scalability
August 5, 2025 at 2:25 PM
PSA: New open access Journal of Systems Scalability launching.
Editor-in-Chief: Carlos Juiz
www.icck.org/jss
#systems #architecture #performance #scalability
Editor-in-Chief: Carlos Juiz
www.icck.org/jss
#systems #architecture #performance #scalability
GCAP Day 10. What you missed:
❥ Avalanche vs. Paxos coordination scalability
❥ Memcache scalability analysis with USL
❥ Queueing models of Kubernetes control and data planes
❥ Class dismissed 📚 🎓🍻
❥ Avalanche vs. Paxos coordination scalability
❥ Memcache scalability analysis with USL
❥ Queueing models of Kubernetes control and data planes
❥ Class dismissed 📚 🎓🍻
August 1, 2025 at 10:43 PM
GCAP Day 10. What you missed:
❥ Avalanche vs. Paxos coordination scalability
❥ Memcache scalability analysis with USL
❥ Queueing models of Kubernetes control and data planes
❥ Class dismissed 📚 🎓🍻
❥ Avalanche vs. Paxos coordination scalability
❥ Memcache scalability analysis with USL
❥ Queueing models of Kubernetes control and data planes
❥ Class dismissed 📚 🎓🍻
Never let reality get in the way of a good performance model.
July 31, 2025 at 11:35 PM
Never let reality get in the way of a good performance model.
GCAP Day 9. What you missed:
❥ Rate limiting services via Guerrilla queueing models
❥ Performance visualization tools
❥ Zookeeper scalability analysis with the USL
www.perfdynamics.com/Manifesto/US...
❥ Rate limiting services via Guerrilla queueing models
❥ Performance visualization tools
❥ Zookeeper scalability analysis with the USL
www.perfdynamics.com/Manifesto/US...
July 31, 2025 at 8:21 PM
GCAP Day 9. What you missed:
❥ Rate limiting services via Guerrilla queueing models
❥ Performance visualization tools
❥ Zookeeper scalability analysis with the USL
www.perfdynamics.com/Manifesto/US...
❥ Rate limiting services via Guerrilla queueing models
❥ Performance visualization tools
❥ Zookeeper scalability analysis with the USL
www.perfdynamics.com/Manifesto/US...
GCAP Day 8. What you missed:
❥ Linear vs Dynamic programming
❥ Benchmarking/load testing prats and pitfalls
❥ Prelim discussion about Kefka, Kubernetes
❥ What kind of scheduler is in the K8s control plane?
❥ Linear vs Dynamic programming
❥ Benchmarking/load testing prats and pitfalls
❥ Prelim discussion about Kefka, Kubernetes
❥ What kind of scheduler is in the K8s control plane?
July 30, 2025 at 10:17 PM
GCAP Day 8. What you missed:
❥ Linear vs Dynamic programming
❥ Benchmarking/load testing prats and pitfalls
❥ Prelim discussion about Kefka, Kubernetes
❥ What kind of scheduler is in the K8s control plane?
❥ Linear vs Dynamic programming
❥ Benchmarking/load testing prats and pitfalls
❥ Prelim discussion about Kefka, Kubernetes
❥ What kind of scheduler is in the K8s control plane?
GCAP Day 7. What you missed:
❥ Python least_squares() seems to be a bloated implementation of codes in the "Numerical Recipes" book.
❥ To apply the USL, use curve_fit() instead.
❥ Multivariate regression: How to reattach (noisy) fingers to amputated daily "numero uno" data
❥ Python least_squares() seems to be a bloated implementation of codes in the "Numerical Recipes" book.
❥ To apply the USL, use curve_fit() instead.
❥ Multivariate regression: How to reattach (noisy) fingers to amputated daily "numero uno" data
July 29, 2025 at 8:25 PM
GCAP Day 7. What you missed:
❥ Python least_squares() seems to be a bloated implementation of codes in the "Numerical Recipes" book.
❥ To apply the USL, use curve_fit() instead.
❥ Multivariate regression: How to reattach (noisy) fingers to amputated daily "numero uno" data
❥ Python least_squares() seems to be a bloated implementation of codes in the "Numerical Recipes" book.
❥ To apply the USL, use curve_fit() instead.
❥ Multivariate regression: How to reattach (noisy) fingers to amputated daily "numero uno" data
GCAP Day 6. What you missed:
❥ USL scalability analysis
lnkd.in/gxaaAiww
❥ Student gets PDQ 7.0 running on Python 3.12 👏🎉
lnkd.in/gPzzHiez
❥ Homework Q-5 solved (a) analytically (b) cf. with corresponding PyDQ model
- Discussed why analytic calculation was wrong
❥ USL scalability analysis
lnkd.in/gxaaAiww
❥ Student gets PDQ 7.0 running on Python 3.12 👏🎉
lnkd.in/gPzzHiez
❥ Homework Q-5 solved (a) analytically (b) cf. with corresponding PyDQ model
- Discussed why analytic calculation was wrong
July 28, 2025 at 10:31 PM
GCAP Day 6. What you missed:
❥ USL scalability analysis
lnkd.in/gxaaAiww
❥ Student gets PDQ 7.0 running on Python 3.12 👏🎉
lnkd.in/gPzzHiez
❥ Homework Q-5 solved (a) analytically (b) cf. with corresponding PyDQ model
- Discussed why analytic calculation was wrong
❥ USL scalability analysis
lnkd.in/gxaaAiww
❥ Student gets PDQ 7.0 running on Python 3.12 👏🎉
lnkd.in/gPzzHiez
❥ Homework Q-5 solved (a) analytically (b) cf. with corresponding PyDQ model
- Discussed why analytic calculation was wrong
Ditto for the performance analysis & capacity planning of computer systems in general:
- Assume failure
- Plan for resilience
- Monitor everything
- Analyze and predict performance and capacity
- Have an exceptional performance team of skilled experts
lnkd.in/g8eGRnYq
- Assume failure
- Plan for resilience
- Monitor everything
- Analyze and predict performance and capacity
- Have an exceptional performance team of skilled experts
lnkd.in/g8eGRnYq
July 27, 2025 at 4:41 PM
Ditto for the performance analysis & capacity planning of computer systems in general:
- Assume failure
- Plan for resilience
- Monitor everything
- Analyze and predict performance and capacity
- Have an exceptional performance team of skilled experts
lnkd.in/g8eGRnYq
- Assume failure
- Plan for resilience
- Monitor everything
- Analyze and predict performance and capacity
- Have an exceptional performance team of skilled experts
lnkd.in/g8eGRnYq
I recently discovered the original 1992 paper where I first introduced the concept of barycentric coordinates for visualizing multiprocessor/multicore performance. www.perfdynamics.com/papers.pdf
More modern renditions (animations) can be viewed here www.perfdynamics.com/Tools/tools....
More modern renditions (animations) can be viewed here www.perfdynamics.com/Tools/tools....
July 1, 2025 at 6:03 PM
I recently discovered the original 1992 paper where I first introduced the concept of barycentric coordinates for visualizing multiprocessor/multicore performance. www.perfdynamics.com/papers.pdf
More modern renditions (animations) can be viewed here www.perfdynamics.com/Tools/tools....
More modern renditions (animations) can be viewed here www.perfdynamics.com/Tools/tools....
Guerrilla training classes (GCAP) online are brutal.
Despite the rumors, there are no long lunches.
Only performance zealots should apply.
Next class starts July 21, 2025.
www.perfdynamics.com/Classes/sche...
#Performance #Training
Despite the rumors, there are no long lunches.
Only performance zealots should apply.
Next class starts July 21, 2025.
www.perfdynamics.com/Classes/sche...
#Performance #Training
June 27, 2025 at 2:05 PM
Guerrilla training classes (GCAP) online are brutal.
Despite the rumors, there are no long lunches.
Only performance zealots should apply.
Next class starts July 21, 2025.
www.perfdynamics.com/Classes/sche...
#Performance #Training
Despite the rumors, there are no long lunches.
Only performance zealots should apply.
Next class starts July 21, 2025.
www.perfdynamics.com/Classes/sche...
#Performance #Training
Guerrilla Capacity Planning (GCAP) Online
— EU Edition — registration is now OPEN
Dates: July 21 through August 1, 2025
Times: 1500 - 1900 CET
www.perfdynamics.com/Classes/sche...
Click on the DETAILS box (at the RHS) to register
www.eventbrite.com/e/guerrilla-...
— EU Edition — registration is now OPEN
Dates: July 21 through August 1, 2025
Times: 1500 - 1900 CET
www.perfdynamics.com/Classes/sche...
Click on the DETAILS box (at the RHS) to register
www.eventbrite.com/e/guerrilla-...
June 24, 2025 at 4:22 PM
Guerrilla Capacity Planning (GCAP) Online
— EU Edition — registration is now OPEN
Dates: July 21 through August 1, 2025
Times: 1500 - 1900 CET
www.perfdynamics.com/Classes/sche...
Click on the DETAILS box (at the RHS) to register
www.eventbrite.com/e/guerrilla-...
— EU Edition — registration is now OPEN
Dates: July 21 through August 1, 2025
Times: 1500 - 1900 CET
www.perfdynamics.com/Classes/sche...
Click on the DETAILS box (at the RHS) to register
www.eventbrite.com/e/guerrilla-...
Thank you @Lyft for great (honest) drivers and great (fast) customer service. 👍
I'll continue to be a Return customer ... with my phone.
I'll continue to be a Return customer ... with my phone.
June 20, 2025 at 8:53 PM
Thank you @Lyft for great (honest) drivers and great (fast) customer service. 👍
I'll continue to be a Return customer ... with my phone.
I'll continue to be a Return customer ... with my phone.
Bill Atkinson, he of QuickDraw and MacPaint fame, has died at 74 of pancreatic cancer. news.ycombinator.com/item?id=4421...
June 8, 2025 at 2:01 AM
Bill Atkinson, he of QuickDraw and MacPaint fame, has died at 74 of pancreatic cancer. news.ycombinator.com/item?id=4421...
Interesting to see LoadRunner is still alive. I haven't heard about LR anything in a long time, let alone LR Cloud. #load #test #performance www.linkedin.com/pulse/loadru...
June 5, 2025 at 4:04 PM
Interesting to see LoadRunner is still alive. I haven't heard about LR anything in a long time, let alone LR Cloud. #load #test #performance www.linkedin.com/pulse/loadru...
Best Boggle board found after 20 years of casual computer searching
www.bbc.co.uk/sounds/play/... starting @ 19:08 - Peak word: "replastering"
www.bbc.co.uk/sounds/play/... starting @ 19:08 - Peak word: "replastering"
May 24, 2025 at 11:44 PM
Best Boggle board found after 20 years of casual computer searching
www.bbc.co.uk/sounds/play/... starting @ 19:08 - Peak word: "replastering"
www.bbc.co.uk/sounds/play/... starting @ 19:08 - Peak word: "replastering"
"DeepSeek-V3: Scaling Challenges," in Proc. 52nd Intl. Symp. Comp. Arch. (ISCA 2025), June 21-25, Tokyo, Japan. #scalability #Performance #GenAI #AI
arxiv.org/pdf/2505.093...
arxiv.org/pdf/2505.093...
May 24, 2025 at 4:17 AM
"DeepSeek-V3: Scaling Challenges," in Proc. 52nd Intl. Symp. Comp. Arch. (ISCA 2025), June 21-25, Tokyo, Japan. #scalability #Performance #GenAI #AI
arxiv.org/pdf/2505.093...
arxiv.org/pdf/2505.093...