ldr: odt has always seemingly been used?

This commit is contained in:
Lightos1 committed 2026-10-16 20:55:06 +02:00
1 parent dbc37c6982
commit b1b4ead2e2
6 files changed
+11 -10

No files matched your search

@@ -56,10 +56,8 @@ namespace ams::ldr::hoc::pcv::erista {
double fullRoundDelay = flyByTime + tDQSCK_max;
/* Read to write turnaround. */
/* Nvidia seem to use a 0.25 constant here for tK1, I am 99% certain this is tDQSTA. */
/* This still left a mismatch; after some brute forcing I got GET_CYCLE_FLOOR(1.5) - 8 (using a static latency) as the cleanest result. */
/* The latency definitely is not WL, but when testing for every latency, it lines up perfectly with wqsOn and wl seems to cancel out? */
tR2W = RL + (BL / 2) + CEIL((fullRoundDelay / tCK_avg) + tDQSTA) - GET_CYCLE_FLOOR(1.5) - GetWdqsOn() + tWPRE - (C.t6_tRTW * 3) + finetRTW;
/* Switch seemingly uses odt. */
tR2W = RL + (BL / 2) + CEIL((fullRoundDelay / tCK_avg) + tDQSTA) - GET_CYCLE_FLOOR(tODTon) - GetOdtLon() + tWPRE - (C.t6_tRTW * 3) + finetRTW;
/* Read data valid: Read command -> data ready to be latched into the registers. */
rdv = RL + GET_CYCLE_CEIL(fullRoundDelay) + 18;