cinera_handmade.network/miotatsu/riscy/book/coad011.hmml

85 lines
6.2 KiB
Plaintext
Raw Normal View History

[video member=miotatsu stream_platform=twitch project=book title="1.13 Exercises 9-10" vod_platform=youtube id=JgxH7lkETqc annotator=Miblo]
[0:06][Recap and set the stage for the day, with no internet]
[5:08][Chapter 1.13, Exercise 1.9 - Parallelism[ref
title="Computer Organization and Design RISC-V Edition"
author="David Patterson & John Hennessy"
publisher="Morgan Kaufmann"
isbn=9780128122754
url=https://www.elsevier.com/books/computer-organization-and-design-risc-v-edition/patterson/978-0-12-812275-4]][:performance :research]
[6:37][Chapter 1.13, Exercise 1.9 - CPIs for our instructions][:blackboard]
[12:21][Chapter 1.13, Exercise 1.9.1 - Total execution time and speedup over multiple cores[ref
title="Computer Organization and Design RISC-V Edition"
author="David Patterson & John Hennessy"
publisher="Morgan Kaufmann"
isbn=9780128122754
url=https://www.elsevier.com/books/computer-organization-and-design-risc-v-edition/patterson/978-0-12-812275-4]][:performance :research]
[14:46][Chapter 1.13, Exercise 1.9.1 - Computing that execution time and speedup over multiple cores[ref
site=Wikipedia
page=Hertz
url=https://en.wikipedia.org/wiki/Hertz]][:blackboard :performance]
[23:41][Consider parallelising functions][:optimisation]
[27:18][@croepha][Hello]
[27:39][Chapter 1.13, Exercise 1.9.1 continued][:blackboard :performance]
[30:57][Consider the number of characters in this computation, for pcalc]
[32:35][Chapter 1.13, Exercise 1.9.1 continued][:blackboard :performance]
[35:32][@croepha][I try to bypass X11 as much as possible. I pretty much just get an OpenGL context with glx and then do direct rendering][:rendering]
[38:05][Recommend BetterOS.org[ref
site=Twitter
page="I stumbled onto a real gem today: \"Low-Level Graphics on Linux\""
url=https://twitter.com/hmn_riscy/status/895532169891790848]]
[39:40][Chapter 1.13, Exercise 1.9.1 continued][:blackboard :performance]
[40:39][Plug the 'COAD 1.13 Exercises' forum thread[ref
site="RISCY BUSINESS Forums"
page="COAD 1.13 Exercises"
url=https://riscy.handmade.network/forums/t/2546-coad_1.13_exercises]]
[41:13][Chapter 1.13, Exercise 1.9.2 - Impact of doubling the CPI of the arithmetic instruction[ref
title="Computer Organization and Design RISC-V Edition"
author="David Patterson & John Hennessy"
publisher="Morgan Kaufmann"
isbn=9780128122754
url=https://www.elsevier.com/books/computer-organization-and-design-risc-v-edition/patterson/978-0-12-812275-4]][:performance :research]
[42:17][Chapter 1.13, Exercise 1.9.2 - Computing the impact of doubling the arithmetic instruction CPI][:blackboard :performance]
[46:33][@croepha][Do you find it important to do the math out by hand? I'd be really tempted to just use a spreadsheet]
[48:12][Chapter 1.13, Exercise 1.9.2 continued][:blackboard :performance]
[48:34][Chapter 1.13, Exercise 1.9.3 - Reducing the CPI of load / store instructions to enable a single processor to match a quad core[ref
title="Computer Organization and Design RISC-V Edition"
author="David Patterson & John Hennessy"
publisher="Morgan Kaufmann"
isbn=9780128122754
url=https://www.elsevier.com/books/computer-organization-and-design-risc-v-edition/patterson/978-0-12-812275-4]][:performance :research]
[49:38][Chapter 1.13, Exercise 1.9.3 - Computing the necessary CPI reduction][:blackboard :performance]
[55:01][@insofaras][I also got 3.8]
[55:23][Chapter 1.13, Exercise 1.9.3 continued][:blackboard :performance]
[55:44][@insofaras][But I just tuned in and don't know what I'm doing]
[56:30][Chapter 1.13, Exercise 1.9.3 continued][:blackboard :performance]
[1:00:00][Chapter 1.13, Exercise 1.10 - :Fabrication yield and cost[ref
title="Computer Organization and Design RISC-V Edition"
author="David Patterson & John Hennessy"
publisher="Morgan Kaufmann"
isbn=9780128122754
url=https://www.elsevier.com/books/computer-organization-and-design-risc-v-edition/patterson/978-0-12-812275-4]][:research]
[1:03:12][Chapter 1.13, Exercise 1.10.1 - Computing the yield for our wafers][:blackboard :fabrication]
[1:05:35][@insofaras][Are these exercises available online?]
[1:07:14][Tease a giveaway for reaching 100 YouTube subscribers[ref
site=YouTube
page="RISCY BUSINESS"
url=https://www.youtube.com/channel/UCDw2ohIbGAoWWMq8fVAfNXg]]
[1:08:21][@croepha][Might be worth contacting the publisher. They might just give you some stuff, as it might be good PR for them]
[1:10:46][@croepha][Hmm, interesting. I think that if you are only using, like, 10% (or something like that) of the original work, then it's considered fair use. But I'm not a lawyer]
[1:13:18][@insofaras][RMS said it was better to steal bread than write proprietary software]
[1:13:35][Chapter 1.13, Exercise 1.10.1 continued][:blackboard :fabrication]
[1:15:47][@croepha][I'm not sure I agree that piracy is a good thing, but we can save that discussion for another time. Don't want to hijack the stream. I could talk about it on my stream some time if someone brings it up]
[1:17:11][Chapter 1.13, Exercise 1.10.1 continued][:blackboard :fabrication]
[1:23:26][@insofaras][I would've thought they'd use square or rectangular dies, but I have, like, no knowledge of this stuff]
[1:25:03][@insofaras][Oh right, maybe I meant wafer]
[1:27:01][Chapter 1.13, Exercise 1.10.1 continued][:blackboard :fabrication]
[1:37:10][Chapter 1.13, Exercise 1.10.2 - Computing the cost per die for both wafers][:blackboard :fabrication]
[1:40:20][Chapter 1.13, Exercise 1.10.3 - Computing the new die area and yield for 10% more dies per yield and 15% more defects per area unit][:blackboard :fabrication]
[1:51:58][Chapter 1.13, Exercise 1.10.1 - Recomputing the yield for our wafers, trusting the algebra][:blackboard :fabrication]
[1:54:45][Chapter 1.13, Exercise 1.10.3 continued][:blackboard :fabrication]
[1:58:04][Chapter 1.13, Exercise 1.10.1 - Finish recomputing the yield for our original wafers][:blackboard :fabrication]
[2:00:20][Chapter 1.13, Exercise 1.10.1 - Recomputing the cost per die for our wafers][:blackboard :fabrication]
[2:01:49][Chapter 1.13, Exercise 1.10.4 - Computing the defects per area unit for a 0.92 to 0.95 increased yield from a die of area 200mm²][:blackboard :fabrication]
[2:14:04][Call it here]
[/video]