Closed
[Bug-Report|缺陷反馈]: aclnnXLogYScalarSelf 声明 self 支持 DOUBLE,实现强降 fp32:|self|>fp32_max 时 1e300*log(1) 返 nan(真值 0) #4296
liuhanwen8Add on 18 days agoClosed at 18 hours ago
18 days ago Added label: bug-report
陈思
18 days ago Comments:
18 days ago Comments:
您好,感谢反馈!
核查代码基线为 cann/ops-math 8c1f633e3679f83b55a237c552f842cdcf80462f;设备现象依据您提供的实测。
问题分析
核查 ScalarSelf实现,self白名单包含DOUBLE,但 CheckPromoteType 将推导出的DOUBLE计算类型转换为FLOAT。随后self经 ConvertToTensor、other经Cast进入该类型,执行XLogY后再转换到out类型。
因此这条路径没有保留DOUBLE计算域。self=1e300转换为fp32时超出有限表示范围;即使other=1使log项为0,也存在 inf*0 产生NaN的路径,这能解释您报告的两段均成功而输出NaN。最终out使用DOUBLE也无法恢复先前转换损失的信息。
处理方向
若保留DOUBLE计算语义,需要核对底层计算支持、类型推导和结果转换,不能只删除一处DOUBLE→FLOAT;若接口定位为FLOAT计算,则应明确声明DOUBLE标量的转换方式及超范围处理约束。直接删除DOUBLE输入支持会影响已有调用,也需评估兼容性。当前建议避免依赖超过fp32有限范围的标量执行此路径。我们会继续分析输入声明与计算域如何保持一致。


18 days ago Assign chen_0715 as responsible person
18 days ago Added label: pending-confirmation
15 days ago Removed label: pending-confirmation
18 hours ago Close issue
18 hours ago changed custom state from 进行中 to 已完成
16 hours ago Added label: resolved
现象
资料
math/xlogy/docs/aclnnXLogYScalarSelf.md:48声明 self 支持 DOUBLE;实测aclScalar取ACL_DOUBLE时入参被接受(ret=0),但实现把推导类型从 DOUBLE 强降为 FLOAT,self 先按 fp32 转换,超出 fp32 表示范围即溢出为 ±inf,再与log(other)相乘得出 nan / ±inf。分歧只出现在
other == 1(真值精确为 0)这类乘数为 0 的位置:inf * 0被判为 nan。根因(
math/xlogy/op_api/aclnn_x_log_y_scalar_self.cpp):promoteType = IsFloatingType(promoteType) ? promoteType : DT_FLOAT;之后紧跟promoteType = (promoteType == DT_DOUBLE) ? DT_FLOAT : promoteType;—— DOUBLE 被无条件降为 FLOAT;ConvertToTensor(self, promoteType)—— self 先转 fp32,1e300 即 +inf。同族
aclnnXLogYScalarOther的标量白名单同样含 DOUBLE,代码路径一致。期望
两种修法任选其一,都要求「声明与行为一致」:
1e300 * log(1.0)返回 0;self的 dtype 列表应删去 DOUBLE,或明确写「DOUBLE 会先转换为 FLOAT 计算」。环境
ASCEND_HOME_PATH=<CANN 安装根>/cann-9.2.0-std=c++17)584b2ea6cbaeba684a51a74296d82d4d50ab3b0f(2026-10-24 提交,本地 clone 未改动);目标文件math/xlogy/复现步骤
repro.cpp;g++ -std=c++17 repro.cpp -o repro -I$ASCEND_HOME_PATH/include -I$ASCEND_HOME_PATH/include/aclnnop -L$ASCEND_HOME_PATH/lib64 -lascendcl -lnnopbase -lopapi -Wl,-rpath,$ASCEND_HOME_PATH/lib64./reproout[0]=0.000000;实测输出out[0]=nan复现脚本
#include <cstdio> #include <vector> #include "acl/acl.h" #include "aclnnop/aclnn_x_log_y_scalar_self.h" int main() { if (aclInit(nullptr) != ACL_SUCCESS) { printf("aclInit failed\n"); return 1; } if (aclrtSetDevice(0) != ACL_SUCCESS) { printf("setDevice failed\n"); return 1; } aclrtStream stream; aclrtCreateStream(&stream); const std::vector<int64_t> shape = {2}; std::vector<float> otherHost = {1.0f, 2.0f}; std::vector<float> outHost(2, 0.0f); void *otherDev = nullptr, *outDev = nullptr; aclrtMalloc(&otherDev, 2 * sizeof(float), ACL_MEM_MALLOC_HUGE_FIRST); aclrtMalloc(&outDev, 2 * sizeof(float), ACL_MEM_MALLOC_HUGE_FIRST); aclrtMemcpy(otherDev, 2 * sizeof(float), otherHost.data(), 2 * sizeof(float), ACL_MEMCPY_HOST_TO_DEVICE); aclrtMemcpy(outDev, 2 * sizeof(float), outHost.data(), 2 * sizeof(float), ACL_MEMCPY_HOST_TO_DEVICE); std::vector<int64_t> strides = {1}; aclTensor* other = aclCreateTensor(shape.data(), 1, ACL_FLOAT, strides.data(), 0, ACL_FORMAT_ND, shape.data(), 1, otherDev); aclTensor* out = aclCreateTensor(shape.data(), 1, ACL_FLOAT, strides.data(), 0, ACL_FORMAT_ND, shape.data(), 1, outDev); double selfValue = 1e300; // self 的 dtype 是 ACL_DOUBLE aclScalar* self = aclCreateScalar(&selfValue, ACL_DOUBLE); uint64_t ws = 0; aclOpExecutor* exec = nullptr; auto ret1 = aclnnXLogYScalarSelfGetWorkspaceSize(self, other, out, &ws, &exec); int ret2 = -1; if (ret1 == 0) { void* wsp = nullptr; if (ws > 0) aclrtMalloc(&wsp, ws, ACL_MEM_MALLOC_HUGE_FIRST); ret2 = (int)aclnnXLogYScalarSelf(wsp, ws, exec, stream); aclrtSynchronizeStream(stream); if (wsp) aclrtFree(wsp); } aclrtMemcpy(outHost.data(), 2 * sizeof(float), outDev, 2 * sizeof(float), ACL_MEMCPY_DEVICE_TO_HOST); printf("ret1=%d ret2=%d workspaceSize=%llu\n", (int)ret1, ret2, (unsigned long long)ws); printf("out[0]=%f (expect 0)\n", outHost[0]); printf("out[1]=%f (expect inf: 6.9e299 overflows fp32)\n", outHost[1]); aclDestroyScalar(self); aclDestroyTensor(other); aclDestroyTensor(out); aclrtFree(otherDev); aclrtFree(outDev); aclrtDestroyStream(stream); aclrtResetDevice(0); aclFinalize(); return 0; }实测输出:
影响
资料声明 DOUBLE 的意义完全落在 fp32 表示不了的区间(|self| > 3.4028e38)——该区间内的 DOUBLE 与 FLOAT 逐位等价,声明与否无差别;而正是这一段给出了错误结果:
x * log(1)这类数学上精确为 0 的表达式返回 nan,x * log(2)类提前溢出为 ±inf。使用 DOUBLE 标量做缩放(如1e39 * log(y))的调用方会拿到全 nan 输出,且没有任何错误码提示。建议
优先按实现侧修(保留 DOUBLE 计算域);若确定只在 fp32 域计算,请同步把
self的 dtype 列表中的 DOUBLE 去掉,或在self使用说明里补一句「DOUBLE 会转换为 FLOAT 参与计算」。同族aclnnXLogYScalarOther/aclnnXLogYTensor有相同代码形态,一并核对。