| b.liu | e958203 | 2025-04-17 19:18:16 +0800 | [diff] [blame^] | 1 | From: Felix Fietkau <nbd@nbd.name> |
| 2 | Date: Fri, 15 Jan 2021 12:15:06 +0100 |
| 3 | Subject: [PATCH] mac80211: minstrel_ht: fix rounding error in throughput |
| 4 | calculation |
| 5 | |
| 6 | On lower data rates, the throughput calculation has a significant rounding |
| 7 | error, causing rates like 48M and 54M OFDM to share the same throughput |
| 8 | value with >= 90% success probablity. |
| 9 | |
| 10 | This is because the result of the division (prob_avg * 1000) / nsecs |
| 11 | is really small (8 in this example). |
| 12 | |
| 13 | Improve accuracy by moving over some zeroes, making better use of the full |
| 14 | range of u32 before the division. |
| 15 | |
| 16 | Signed-off-by: Felix Fietkau <nbd@nbd.name> |
| 17 | --- |
| 18 | |
| 19 | --- a/net/mac80211/rc80211_minstrel_ht.c |
| 20 | +++ b/net/mac80211/rc80211_minstrel_ht.c |
| 21 | @@ -445,10 +445,9 @@ minstrel_ht_get_tp_avg(struct minstrel_h |
| 22 | * (prob is scaled - see MINSTREL_FRAC above) |
| 23 | */ |
| 24 | if (prob_avg > MINSTREL_FRAC(90, 100)) |
| 25 | - return MINSTREL_TRUNC(100000 * ((MINSTREL_FRAC(90, 100) * 1000) |
| 26 | - / nsecs)); |
| 27 | - else |
| 28 | - return MINSTREL_TRUNC(100000 * ((prob_avg * 1000) / nsecs)); |
| 29 | + prob_avg = MINSTREL_FRAC(90, 100); |
| 30 | + |
| 31 | + return MINSTREL_TRUNC(100 * ((prob_avg * 1000000) / nsecs)); |
| 32 | } |
| 33 | |
| 34 | /* |