Timeline for How reassuring is 64-bit (in)security?
Current License: CC BY-SA 4.0
26 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 30, 2022 at 10:58 | history | edited | kelalaka | CC BY-SA 4.0 |
Table has converted to new table format.
|
| Mar 1, 2021 at 14:16 | history | edited | fgrieu♦ | CC BY-SA 4.0 |
Minor fixes
|
| Mar 1, 2021 at 14:11 | history | edited | fgrieu♦ | CC BY-SA 4.0 |
Minor fixes
|
| Mar 1, 2021 at 11:36 | history | edited | kelalaka | CC BY-SA 4.0 |
Sagemath formulas mangled! Formulas corrected. Thanks to FGrieu
|
| Feb 9, 2021 at 13:49 | history | edited | kelalaka | CC BY-SA 4.0 |
And the miners reached 2^93 per year.
|
| Oct 17, 2020 at 15:46 | history | edited | kelalaka | CC BY-SA 4.0 |
added common Nvidia's devices.
|
| Oct 1, 2020 at 9:50 | comment | added | kelalaka | @tleb It should be more exact now. Also, the table extended. The approximations are only performed at the end. Still, the are approximations since I cannot access those machines to see the real performance. | |
| Oct 1, 2020 at 9:48 | history | edited | kelalaka | CC BY-SA 4.0 |
table corrected, extended and the calculations are approximated at the end. Thank for tleb to notice that.
|
| Oct 1, 2020 at 8:06 | comment | added | tleb | Is it $2^{63}$ per day or per hour for Summit? The table and the text contradict. Assuming it's per day, the estimation per year has to be corrected too. | |
| Jul 2, 2020 at 12:55 | history | edited | kelalaka | CC BY-SA 4.0 |
Bitcoin miner soon can reach 2^93 in a year.
|
| Dec 28, 2019 at 19:44 | history | edited | kelalaka | CC BY-SA 4.0 |
table.
|
| Aug 12, 2019 at 19:47 | history | edited | kelalaka | CC BY-SA 4.0 |
update on bitcoin's new record
|
| Dec 17, 2018 at 12:34 | history | edited | kelalaka | CC BY-SA 4.0 |
added summit result and tl;dr; power cost need to check
|
| Nov 26, 2018 at 18:56 | comment | added | kelalaka | @JonathanCross If you are talking about generating SHA-1 hash and using only 64 bits than generic collision is $2^{32}$ with a pc you can achieve. The freedom in Google case id PDF's structure. SHattered is not practicable for every data type. | |
| Nov 26, 2018 at 18:48 | comment | added | Jonathan Cross | @kelalaka 2⁸⁰ would be for a full SHA1 hash collision, but I'm only talking about the lower parts (last 16 hex chars) of the hash. Still, it seems the "freedom" you are talking about would only come from the underlying data (OpenPGP pubkey in this case) so making a usable key would require extra steps of generating new private key => pubkey before hashing. I guess SHAttered isn't particularly applicable here. | |
| Nov 23, 2018 at 20:21 | comment | added | kelalaka | @JonathanCross as Dave commented, it is not clear how one can use the shatter since it requires a freedom. One year is approximately 8765 hours = $2^13$, so Titan may reach $2^75$ in one year but the generic collision is $2^{80}$ | |
| Nov 23, 2018 at 1:32 | comment | added | Jonathan Cross | Would these numbers apply equally to OpenPGP "long" key IDs which are the lower 64 bits of a SHA1 hash of the public key? Related: security.stackexchange.com/questions/198249/… | |
| Nov 2, 2018 at 21:25 | history | edited | kelalaka | CC BY-SA 4.0 |
typo
|
| Nov 1, 2018 at 19:38 | history | edited | kelalaka | CC BY-SA 4.0 |
polish
|
| Oct 31, 2018 at 14:11 | history | edited | kelalaka | CC BY-SA 4.0 |
added security section
|
| Oct 31, 2018 at 13:55 | history | edited | kelalaka | CC BY-SA 4.0 |
used tesla to polish the answer. added initial cost and running electricity cost.
|
| Oct 29, 2018 at 1:50 | comment | added | Gordon Davisson | Depending on the application, you may need to take into account the decrease in this cost over time. A MAC or digital signature that isn't worth forging today might be trivial in 20 years. How long does your security need to last? | |
| Oct 28, 2018 at 13:40 | history | edited | kelalaka | CC BY-SA 4.0 |
correction on calculation; stared 8 GPUS
|
| Oct 28, 2018 at 13:40 | vote | accept | DannyNiu | ||
| Oct 28, 2018 at 13:20 | comment | added | SEJPM | One may want to note that $2^{64}$ SHA1 evaluations cost about 1M USD on AWS using p3.16xlarge instances. | |
| Oct 28, 2018 at 13:16 | history | answered | kelalaka | CC BY-SA 4.0 |