* ntb_perf max chunk order
@ 2024-01-16 7:10 Raju Rangoju
2024-01-16 16:13 ` Dave Jiang
0 siblings, 1 reply; 3+ messages in thread
From: Raju Rangoju @ 2024-01-16 7:10 UTC (permalink / raw)
To: jdmason, dave.jiang, allenbh, ntb; +Cc: Basavaraj Natikar
Hi,
Could you please explain why net_perf does not support larger chunk
sizes? Currently only MAX_CHUNK_ORDER 20 is supported (1M and below)
Thanks,
Raju
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: ntb_perf max chunk order
2024-01-16 7:10 ntb_perf max chunk order Raju Rangoju
@ 2024-01-16 16:13 ` Dave Jiang
2024-01-23 18:49 ` Raju Rangoju
0 siblings, 1 reply; 3+ messages in thread
From: Dave Jiang @ 2024-01-16 16:13 UTC (permalink / raw)
To: Raju Rangoju, jdmason, allenbh, ntb; +Cc: Basavaraj Natikar
On 1/16/24 00:10, Raju Rangoju wrote:
> Hi,
>
> Could you please explain why net_perf does not support larger chunk sizes? Currently only MAX_CHUNK_ORDER 20 is supported (1M and below)
It has to do with the limitation of Linux dma_alloc_coherent() call. Allocating physically (or IOVA) contiguous memory that's extremely large is difficult. In the past, you can do it with limited success up to 8MB with help from IOMMU as I recall. For experimental and testing purposes, you can try larger sizes.
>
> Thanks,
> Raju
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: ntb_perf max chunk order
2024-01-16 16:13 ` Dave Jiang
@ 2024-01-23 18:49 ` Raju Rangoju
0 siblings, 0 replies; 3+ messages in thread
From: Raju Rangoju @ 2024-01-23 18:49 UTC (permalink / raw)
To: Dave Jiang, jdmason, allenbh, ntb; +Cc: Basavaraj Natikar
On 1/16/2024 9:43 PM, Dave Jiang wrote:
>
>
> On 1/16/24 00:10, Raju Rangoju wrote:
>> Hi,
>>
>> Could you please explain why net_perf does not support larger chunk sizes? Currently only MAX_CHUNK_ORDER 20 is supported (1M and below)
>
> It has to do with the limitation of Linux dma_alloc_coherent() call. Allocating physically (or IOVA) contiguous memory that's extremely large is difficult. In the past, you can do it with limited success up to 8MB with help from IOMMU as I recall. For experimental and testing purposes, you can try larger sizes.
Sure Dave, I'll explore that. Thank you.
>>
>> Thanks,
>> Raju
>>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-01-23 18:49 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-01-16 7:10 ntb_perf max chunk order Raju Rangoju
2024-01-16 16:13 ` Dave Jiang
2024-01-23 18:49 ` Raju Rangoju
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).