Proactive history read adaptation without test read

1 . A method for reading data from memory, comprising:

performing read operations on the memory while recording as read retry (RR) biases a set of history entries of voltages used to read the memory;

selecting from the set of history entries the RR biases for subsequent read operations of the memory without performing a test operation with any of the RR biases before the subsequent read operations;

in the subsequent read operations, reading the memory with one or more of the RR biases from the history entries;

when a failed bit count is higher than a threshold for the subsequent read operations, evicting a worst entry based on the failed bit count from the history entries and introducing a new history entry into the set of history entries; and

updating the RR biases based on the new history entry.

2 . The method of claim 1 , further comprising recording the set of history reads per die, or per block, or per word line group.

3 . The method of claim 1 , further comprising randomly selecting the RR biases from the set of history reads to read the memory in the subsequent read operations.

4 . The method of claim 1 , further comprising selecting the RR biases from the set of history reads used to read the memory in the subsequent read operations such that each of the RR biases has an equal chance of usage.

5 . The method of claim 1 , further comprising determining the new history entry based on a gradient of differences between voltage thresholds used to read a page of the memory for the history entry having a lowest failed bit count (FBC).

6 . The method of claim 5 , wherein the memory comprises a triple level cell (TLC), and the voltage thresholds are voltages associated with reading one or more of a most significant bit page, a center significant bit page, and a least significant bit page.

7 . The method of claim 6 , wherein the voltage thresholds are voltages V2, V6 for reading the least significant bit page.

8 . The method of claim 7 , further comprising determining the new history entry as (V2−delta*dFBC/dV2, V6−delta*dFBC/dV6), where dFBC/dV2 and dFBC/dV6 are numerical gradients in failed bit counts (FBCs) at the voltage thresholds V2 and V6, and delta is a constant defining a stepping size of the gradients.

9 . The method of claim 8 , wherein the numerical gradients are calculated from differences in failed bit counts for data read at respective voltages offset from voltage thresholds V2 and V6.

10 . The method of claim 5 , further comprising performing the test operation of one or more the RR biases after the updating of the RR biases to adjust the voltages in the set of the history entries.

11 . A memory system, comprising:

a memory;

a controller coupled to the memory and configured to read data from the memory, wherein the controller is configured to:

perform read operations on the memory while recording as read retry (RR) biases a set of history entries of voltages used to read the memory;

select from the set of history entries the RR biases for subsequent read operations of the memory without performing a test operation with any of the RR biases before the subsequent read operations;

in the subsequent read operations, read the memory with one or more of the RR biases from the history entries;

when a failed bit count is higher than a threshold for the subsequent read operations, evict a worst entry based on the failed bit count from the history entries and introduce a new history entry into the set of history entries; and

update the RR biases based on the new history entry.

12 . The memory system of claim 11 , wherein the controller is configured to record the set of history reads per die, or per block, or per word line group.

13 . The memory system of claim 11 , wherein the controller is configured to randomly select the RR biases from the set of history reads to read the memory in the subsequent read operations.

14 . The memory system of claim 11 , wherein the controller is configured to select the RR biases from the set of history reads used to read the memory in the subsequent read operations such that each of the RR biases has an equal chance of usage.

15 . The memory system of claim 11 , wherein the controller is configured to determine the new history entry based on a gradient of differences between voltage thresholds used to read a page of the memory for the history entry having a lowest failed bit count (FBC).

16 . The memory system of claim 15 , wherein the memory comprises a triple level cell (TLC), and the voltage thresholds are voltages associated with reading one or more of a most significant bit page, a center significant bit page, and a least significant bit page.

17 . The memory system of claim 16 , wherein the voltage thresholds are voltages V2, V6 for reading the least significant bit page.

18 . The memory system of claim 17 , wherein the new history entry is determined as (V2−delta*dFBC/dV2, V6−delta*dFBC/dV6), where dFBC/dV2 and dFBC/dV6 are numerical gradients in failed bit counts (FBCs) at the voltage thresholds V2 and V6, and delta is a constant defining a stepping size of the gradients.

19 . The memory system of claim 18 , wherein the controller is configured to calculate the numerical gradients from differences in failed bit counts for data read at respective voltages offset from voltage thresholds V2 and V6.

20 . The memory system of claim 19 , wherein the controller is configured to perform the test operation of one or more the RR biases after the updating of the RR biases to adjust the voltages in the set of the history entries.