flywithu/ntel-gpu-tools
1
0
Fork 0
mirror of https://github.com/tiagovignatti/intel-gpu-tools.git synced 2025-11-04 03:58:27 +00:00
Code Issues Releases Wiki Activity

benchmarks/gem_latency: Use RCS on Sandybridge

Browse Source 
Reading BCS_TIMESTAMP just returns 0...

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This commit is contained in:
Chris Wilson 2015-12-19 17:05:32 +00:00
parent c0942bf528
commit 27e093dd1f
1 changed files with 19 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
benchmarks/gem_latency.c
View File

@ -44,6 +44,10 @@
static int done; static int done;
static int fd; static int fd;
static volatile uint32_t *timestamp_reg;
#define REG(x) (volatile uint32_t *)((volatile char *)igt_global_mmio + x)
#define REG_OFFSET(x) ((volatile char *)(x) - (volatile char *)igt_global_mmio)
struct consumer { struct consumer {
pthread_t thread; pthread_t thread;
@ -92,8 +96,9 @@ struct producer {
#define WIDTH 1024 #define WIDTH 1024
#define HEIGHT 1024 #define HEIGHT 1024
#define RCS_TIMESTAMP (0x2000 + 0x358)
#define BCS_TIMESTAMP (0x22000 + 0x358) #define BCS_TIMESTAMP (0x22000 + 0x358)
#define CYCLES_TO_NS(x) (80*(x)) #define CYCLES_TO_NS(x) (80.*(x))
#define CYCLES_TO_US(x) (CYCLES_TO_NS(x)/1000.) #define CYCLES_TO_US(x) (CYCLES_TO_NS(x)/1000.)
static uint32_t create_workload(int gen, uint32_t scratch, int factor) static uint32_t create_workload(int gen, uint32_t scratch, int factor)
@ -193,7 +198,7 @@ static void setup_latency(struct producer *p, int gen)
map[i++] = 0x24 << 23 | 1; map[i++] = 0x24 << 23 | 1;
if (has_64bit_reloc) if (has_64bit_reloc)
map[i-1]++; map[i-1]++;
map[i++] = BCS_TIMESTAMP; map[i++] = REG_OFFSET(timestamp_reg);
p->latency_dispatch.reloc[0].offset = i * sizeof(uint32_t); p->latency_dispatch.reloc[0].offset = i * sizeof(uint32_t);
p->latency_dispatch.reloc[0].delta = 4000; p->latency_dispatch.reloc[0].delta = 4000;
p->latency_dispatch.reloc[0].target_handle = handle; p->latency_dispatch.reloc[0].target_handle = handle;
@ -238,11 +243,10 @@ static void setup_nop(struct producer *p, uint32_t batch)
eb->rsvd1 = p->ctx; eb->rsvd1 = p->ctx;
} }
#define READ(x) *(volatile uint32_t *)((volatile char *)igt_global_mmio + x)
static void measure_latency(struct producer *p, igt_stats_t *stats) static void measure_latency(struct producer *p, igt_stats_t *stats)
{ {
gem_sync(fd, p->latency_dispatch.exec[0].handle); gem_sync(fd, p->latency_dispatch.exec[0].handle);
igt_stats_push(stats, READ(BCS_TIMESTAMP) - *p->last_timestamp); igt_stats_push(stats, *timestamp_reg - *p->last_timestamp);
} }
static void *producer(void *arg) static void *producer(void *arg)
@ -251,7 +255,7 @@ static void *producer(void *arg)
int n; int n;
while (!done) { while (!done) {
uint32_t start = READ(BCS_TIMESTAMP); uint32_t start = *timestamp_reg;
int batches; int batches;
/* Control the amount of work we do, similar to submitting /* Control the amount of work we do, similar to submitting
@ -374,6 +378,16 @@ static int run(int seconds,
intel_register_access_init(intel_get_pci_device(), false); intel_register_access_init(intel_get_pci_device(), false);
if (gen == 6)
timestamp_reg = REG(RCS_TIMESTAMP);
else
timestamp_reg = REG(BCS_TIMESTAMP);
nrun = *timestamp_reg;
usleep(1);
if (*timestamp_reg == nrun)
return 77;
scratch = gem_create(fd, 4*WIDTH*HEIGHT); scratch = gem_create(fd, 4*WIDTH*HEIGHT);
nop_batch = create_nop(); nop_batch = create_nop();
workload_batch = create_workload(gen, scratch, workload); workload_batch = create_workload(gen, scratch, workload);