summarylogtreecommitdiffstats
path: root/100-nvidia-open-tinygrad.patch
diff options
context:
space:
mode:
authorJennifer Zhou2024-05-02 04:42:18 +0000
committerJennifer Zhou2024-05-02 04:42:18 +0000
commita719a9e779d1aea03aca01468083bbb78dcf8d45 (patch)
treeb16280dec714109172a9619ed1baba7eb44671ac /100-nvidia-open-tinygrad.patch
parentcc5ccb9f7167fd24fa830be9869b8f15e87244d9 (diff)
downloadaur-a719a9e779d1aea03aca01468083bbb78dcf8d45.tar.gz
update
Diffstat (limited to '100-nvidia-open-tinygrad.patch')
-rw-r--r--100-nvidia-open-tinygrad.patch55
1 files changed, 35 insertions, 20 deletions
diff --git a/100-nvidia-open-tinygrad.patch b/100-nvidia-open-tinygrad.patch
index 69213846cc3b..466a9d8e373d 100644
--- a/100-nvidia-open-tinygrad.patch
+++ b/100-nvidia-open-tinygrad.patch
@@ -1,12 +1,12 @@
diff --git a/README.md b/README.md
-index 77ad719f..d1d386f2 100644
+index 5b154a84..d1d386f2 100644
--- a/README.md
+++ b/README.md
-@@ -1,915 +1,125 @@
+@@ -1,930 +1,125 @@
-# NVIDIA Linux Open GPU Kernel Module Source
-
-This is the source release of the NVIDIA Linux open GPU kernel modules,
--version 550.54.15.
+-version 550.76.
-
-
-## How to Build
@@ -22,7 +22,7 @@ index 77ad719f..d1d386f2 100644
-
-Note that the kernel modules built here must be used with GSP
-firmware and user-space NVIDIA GPU driver components from a corresponding
--550.54.15 driver release. This can be achieved by installing
+-550.76 driver release. This can be achieved by installing
-the NVIDIA GPU driver from the .run file using the `--no-kernel-modules`
-option. E.g.,
-
@@ -193,7 +193,7 @@ index 77ad719f..d1d386f2 100644
-For details on feature support and limitations, see the NVIDIA GPU driver
-end user README here:
-
--https://us.download.nvidia.com/XFree86/Linux-x86_64/550.54.15/README/kernel_open.html
+-https://us.download.nvidia.com/XFree86/Linux-x86_64/550.76/README/kernel_open.html
-
-For vGPU support, please refer to the README.vgpu packaged in the vGPU Host
-Package for more details.
@@ -656,6 +656,7 @@ index 77ad719f..d1d386f2 100644
-| NVIDIA T1000 8GB | 1FF0 17AA 1612 |
-| NVIDIA T400 4GB | 1FF2 1028 1613 |
-| NVIDIA T400 4GB | 1FF2 103C 1613 |
+-| NVIDIA T400E | 1FF2 103C 18FF |
-| NVIDIA T400 4GB | 1FF2 103C 8A80 |
-| NVIDIA T400 4GB | 1FF2 10DE 1613 |
-| NVIDIA T400E | 1FF2 10DE 18FF |
@@ -834,6 +835,14 @@ index 77ad719f..d1d386f2 100644
-| NVIDIA GeForce RTX 3050 4GB Laptop GPU | 25AB |
-| NVIDIA GeForce RTX 3050 6GB Laptop GPU | 25AC |
-| NVIDIA GeForce RTX 2050 | 25AD |
+-| NVIDIA RTX A1000 | 25B0 1028 1878 |
+-| NVIDIA RTX A1000 | 25B0 103C 1878 |
+-| NVIDIA RTX A1000 | 25B0 10DE 1878 |
+-| NVIDIA RTX A1000 | 25B0 17AA 1878 |
+-| NVIDIA RTX A400 | 25B2 1028 1879 |
+-| NVIDIA RTX A400 | 25B2 103C 1879 |
+-| NVIDIA RTX A400 | 25B2 10DE 1879 |
+-| NVIDIA RTX A400 | 25B2 17AA 1879 |
-| NVIDIA A16 | 25B6 10DE 14A9 |
-| NVIDIA A2 | 25B6 10DE 157E |
-| NVIDIA RTX A2000 Laptop GPU | 25B8 |
@@ -872,6 +881,7 @@ index 77ad719f..d1d386f2 100644
-| NVIDIA GeForce RTX 4080 SUPER | 2702 |
-| NVIDIA GeForce RTX 4080 | 2704 |
-| NVIDIA GeForce RTX 4070 Ti SUPER | 2705 |
+-| NVIDIA GeForce RTX 4070 | 2709 |
-| NVIDIA GeForce RTX 4090 Laptop GPU | 2717 |
-| NVIDIA RTX 5000 Ada Generation Laptop GPU | 2730 |
-| NVIDIA GeForce RTX 4090 Laptop GPU | 2757 |
@@ -879,6 +889,7 @@ index 77ad719f..d1d386f2 100644
-| NVIDIA GeForce RTX 4070 Ti | 2782 |
-| NVIDIA GeForce RTX 4070 SUPER | 2783 |
-| NVIDIA GeForce RTX 4070 | 2786 |
+-| NVIDIA GeForce RTX 4060 Ti | 2788 |
-| NVIDIA GeForce RTX 4080 Laptop GPU | 27A0 |
-| NVIDIA RTX 4000 SFF Ada Generation | 27B0 1028 16FA |
-| NVIDIA RTX 4000 SFF Ada Generation | 27B0 103C 16FA |
@@ -901,6 +912,7 @@ index 77ad719f..d1d386f2 100644
-| NVIDIA RTX 3500 Ada Generation Embedded GPU | 27FB |
-| NVIDIA GeForce RTX 4060 Ti | 2803 |
-| NVIDIA GeForce RTX 4060 Ti | 2805 |
+-| NVIDIA GeForce RTX 4060 | 2808 |
-| NVIDIA GeForce RTX 4070 Laptop GPU | 2820 |
-| NVIDIA RTX 3000 Ada Generation Laptop GPU | 2838 |
-| NVIDIA GeForce RTX 4070 Laptop GPU | 2860 |
@@ -909,8 +921,11 @@ index 77ad719f..d1d386f2 100644
-| NVIDIA GeForce RTX 4050 Laptop GPU | 28A1 |
-| NVIDIA RTX 2000 Ada Generation | 28B0 1028 1870 |
-| NVIDIA RTX 2000 Ada Generation | 28B0 103C 1870 |
+-| NVIDIA RTX 2000E Ada Generation | 28B0 103C 1871 |
-| NVIDIA RTX 2000 Ada Generation | 28B0 10DE 1870 |
+-| NVIDIA RTX 2000E Ada Generation | 28B0 10DE 1871 |
-| NVIDIA RTX 2000 Ada Generation | 28B0 17AA 1870 |
+-| NVIDIA RTX 2000E Ada Generation | 28B0 17AA 1871 |
-| NVIDIA RTX 2000 Ada Generation Laptop GPU | 28B8 |
-| NVIDIA RTX 1000 Ada Generation Laptop GPU | 28B9 |
-| NVIDIA RTX 500 Ada Generation Laptop GPU | 28BA |
@@ -1124,7 +1139,7 @@ index 39cc8812..efd4cda8 100644
(data32 <= NV_REG_STR_RM_FORCE_P2P_TYPE_MAX))
{
diff --git a/src/nvidia/src/kernel/gpu/bus/arch/hopper/kern_bus_gh100.c b/src/nvidia/src/kernel/gpu/bus/arch/hopper/kern_bus_gh100.c
-index 17bc6efc..b62a0919 100644
+index dcac5b9e..6423745f 100644
--- a/src/nvidia/src/kernel/gpu/bus/arch/hopper/kern_bus_gh100.c
+++ b/src/nvidia/src/kernel/gpu/bus/arch/hopper/kern_bus_gh100.c
@@ -333,7 +333,7 @@ kbusVerifyBar2_GH100
@@ -1291,10 +1306,10 @@ index d9a04068..8af18d72 100644
return NV_ERR_NOT_SUPPORTED;
diff --git a/src/nvidia/src/kernel/gpu/bus/p2p_api.c b/src/nvidia/src/kernel/gpu/bus/p2p_api.c
-index c98befb0..0dafbf1c 100644
+index 3d6e1bed..59327bba 100644
--- a/src/nvidia/src/kernel/gpu/bus/p2p_api.c
+++ b/src/nvidia/src/kernel/gpu/bus/p2p_api.c
-@@ -567,20 +567,20 @@ p2papiConstruct_IMPL
+@@ -575,20 +575,20 @@ p2papiConstruct_IMPL
//
// TODO: This function need to have a cleanup path when this function
// fails after kbusCreateP2PMaping(), busBindLocalGfidForP2P()
@@ -1318,7 +1333,7 @@ index c98befb0..0dafbf1c 100644
pRemoteKernelBus, &egmPeer1, &egmPeer2,
pP2PApi->attributes |
DRF_DEF(_P2PAPI, _ATTRIBUTES, _REMOTE_EGM, _YES)));
-@@ -590,14 +590,14 @@ p2papiConstruct_IMPL
+@@ -598,14 +598,14 @@ p2papiConstruct_IMPL
(pCallContext->secInfo.privLevel >= RS_PRIV_LEVEL_KERNEL))
{
NV_CHECK_OK_OR_RETURN(LEVEL_ERROR,
@@ -1336,7 +1351,7 @@ index c98befb0..0dafbf1c 100644
&pNv503bAllocParams->p2lBar1P2PDmaInfo.dma_address,
&pNv503bAllocParams->p2lBar1P2PDmaInfo.dma_size));
}
-@@ -732,7 +732,7 @@ p2papiDestruct_IMPL
+@@ -740,7 +740,7 @@ p2papiDestruct_IMPL
{
// remove any resources associated with this mapping
NV_CHECK_OK_OR_GOTO(status, LEVEL_ERROR,
@@ -1345,9 +1360,9 @@ index c98befb0..0dafbf1c 100644
pRemoteGpu, pRemoteKernelBus,
pP2PApi->peerId1, pP2PApi->peerId2,
pP2PApi->attributes), end);
-@@ -740,7 +740,7 @@ p2papiDestruct_IMPL
- memmgrIsLocalEgmEnabled(GPU_GET_MEMORY_MANAGER(pLocalGpu)) &&
- memmgrIsLocalEgmEnabled(GPU_GET_MEMORY_MANAGER(pRemoteGpu)))
+@@ -749,7 +749,7 @@ p2papiDestruct_IMPL
+ memmgrIsLocalEgmEnabled(GPU_GET_MEMORY_MANAGER(pRemoteGpu)) &&
+ !GPU_IS_NVSWITCH_DETECTED(pLocalGpu))
{
- status = kbusRemoveP2PMapping_HAL(pLocalGpu, pLocalKernelBus,
+ status = kbusRemoveP2PMapping_GH100(pLocalGpu, pLocalKernelBus,
@@ -1355,7 +1370,7 @@ index c98befb0..0dafbf1c 100644
pP2PApi->egmPeerId1, pP2PApi->egmPeerId2,
pP2PApi->attributes |
diff --git a/src/nvidia/src/kernel/rmapi/nv_gpu_ops.c b/src/nvidia/src/kernel/rmapi/nv_gpu_ops.c
-index 432e77e7..aee4195b 100644
+index a7d5c134..245ca7c5 100644
--- a/src/nvidia/src/kernel/rmapi/nv_gpu_ops.c
+++ b/src/nvidia/src/kernel/rmapi/nv_gpu_ops.c
@@ -3058,7 +3058,8 @@ nvGpuOpsBuildExternalAllocPtes
@@ -1394,8 +1409,8 @@ index 432e77e7..aee4195b 100644
{
FlaMemory* pFlaMemory = dynamicCast(pMemory, FlaMemory);
nvFieldSet32(&pPteFmt->fldPeerIndex, peerId, pte.v8);
-@@ -3296,7 +3305,7 @@ nvGpuOpsBuildExternalAllocPtes
- fabricBaseAddress = knvlinkGetUniqueFabricBaseAddress(pMemDesc->pGpu, pKernelNvlink);
+@@ -3303,7 +3312,7 @@ nvGpuOpsBuildExternalAllocPtes
+ }
}
}
- }
@@ -1403,7 +1418,7 @@ index 432e77e7..aee4195b 100644
//
// Both memdescGetPhysAddr() and kgmmuEncodePhysAddr() have pretty high overhead.
-@@ -3451,6 +3460,7 @@ NV_STATUS nvGpuOpsGetExternalAllocPtes(struct gpuAddressSpace *vaSpace,
+@@ -3458,6 +3467,7 @@ NV_STATUS nvGpuOpsGetExternalAllocPtes(struct gpuAddressSpace *vaSpace,
Memory *pMemory = NULL;
PMEMORY_DESCRIPTOR pMemDesc = NULL;
OBJGPU *pMappingGpu = NULL;
@@ -1411,7 +1426,7 @@ index 432e77e7..aee4195b 100644
NvU32 peerId = 0;
NvBool isSliSupported = NV_FALSE;
NvBool isPeerSupported = NV_FALSE;
-@@ -3596,6 +3606,8 @@ NV_STATUS nvGpuOpsGetExternalAllocPtes(struct gpuAddressSpace *vaSpace,
+@@ -3603,6 +3613,8 @@ NV_STATUS nvGpuOpsGetExternalAllocPtes(struct gpuAddressSpace *vaSpace,
&peerId);
if (status != NV_OK)
goto freeGpaMemdesc;
@@ -1420,7 +1435,7 @@ index 432e77e7..aee4195b 100644
}
//
-@@ -3673,7 +3685,7 @@ NV_STATUS nvGpuOpsGetExternalAllocPtes(struct gpuAddressSpace *vaSpace,
+@@ -3680,7 +3692,7 @@ NV_STATUS nvGpuOpsGetExternalAllocPtes(struct gpuAddressSpace *vaSpace,
status = nvGpuOpsBuildExternalAllocPtes(pVAS, pMappingGpu, pAdjustedMemDesc, pMemory, offset, size,
isIndirectPeerSupported, isPeerSupported, peerId,
@@ -1429,7 +1444,7 @@ index 432e77e7..aee4195b 100644
freeGpaMemdesc:
if (pAdjustedMemDesc != pMemDesc)
-@@ -9322,7 +9334,7 @@ NV_STATUS nvGpuOpsGetChannelResourcePtes(struct gpuAddressSpace *vaSpace,
+@@ -9329,7 +9341,7 @@ NV_STATUS nvGpuOpsGetChannelResourcePtes(struct gpuAddressSpace *vaSpace,
status = nvGpuOpsBuildExternalAllocPtes(pVAS, pMappingGpu, pMemDesc, NULL,
offset, size, NV_FALSE, NV_FALSE,