Performance tuning and analysis¶
This is the “best performance practices” document for Gramine performance tuning and explanation of possible software and hardware performance bottlenecks and limitations. In this document, we are highlighting only those manifest options relevant for performance benchmarking. Refer to Manifest syntax for a full list.
Note: The below examples were run on a Mehlow machine. The performance numbers in these examples should not be considered representative and serve only illustration purposes.
Enabling per-thread and process-wide SGX stats¶
See also Profiling with perf below for installing
Enable statistics using
sgx.enable_stats = true manifest option. Now your
graminized application correctly reports performance counters. This is useful
when using e.g.
perf stat to collect performance statistics. This manifest
option also forces Gramine to dump SGX-related information on each
thread/process exit. Here is an example:
libos/test/regression$ perf stat gramine-sgx helloworld Hello world (helloworld)! ----- SGX stats for thread 87219 ----- # of EENTERs: 224 # of EEXITs: 192 # of AEXs: 201 # of sync signals: 32 # of async signals: 0 ----- Total SGX stats for process 87219 ----- # of EENTERs: 224 # of EEXITs: 192 # of AEXs: 201 # of sync signals: 32 # of async signals: 0 Performance counter stats for 'gramine-sgx helloworld': 3,568,568,948 cycles 1,072,072,581 instructions 172,308,653 branches
How to read this output:
- You can see that one thread was created, with Linux-host (actual)
TID = 87219. This one thread belongs to one process, so they have the same
TID = PIDand the same statistics. If the test application would have e.g. two threads, then there would be two “SGX stats for thread” outputs, and their values would sum up to the values reported in the final “Total SGX stats for process”.
- There are about 200 EENTERs and EEXITs. These are mostly due to OCALLs: recall that every OCALL requires one EEXIT to exit the enclave and one EENTER to re-enter it again. We can conclude that there are about 200 OCALLs. Also, the number of EENTERs is slightly higher than of EEXITs: this is because of ECALLs. Recall that Gramine performs one ECALL for the whole process and one ECALL per each new thread (and possibly some more ECALLs to reset threads and other bookkeeping). You can consider ECALLs as “no-return” in Gramine: they only contribute to EENTERs and do not increase EEXITs (in reality, ECALLs perform EEXITs but only after the stats are printed).
- Why there are about 200 OCALLs in our trivial HelloWorld example? This is
because even before HelloWorld’s
main()function starts, Gramine and Glibc initialize themselves. For example, Gramine must open and read the manifest file – this requires several OCALLs. Glibc may need to open and load shared libraries – this requires more OCALLs. In general, you may consider 200 OCALLs as the cost of initialization in Gramine.
- Why are there about 200 Asynchronous Exits (AEXs)? Most of these AEXs come from in-enclave exceptions and normal Linux interrupts. In particular, Linux’s scheduler interrupts each CPU every 4ms, or 250 times per second. Since our enclave workload runs for a fraction of a second, maybe 100 AEXs happen due to scheduler interrupts. Another 32 AEXs happen due to “sync signals” – in particular, SGX and Gramine trap-and-emulate the CPUID instruction. Each trap-and-emulate results in one AEX. Finally, the rest 100 AEXs happen due to some other sources of interrupts: enclave page faults or network interrupts.
- What are the 32 sync signals? These are synchronous signals forwarded by the host Linux to Gramine. Synchronous signals are SIGILL (invalid instruction), SIGFPE (invalid floating-point result), SIGSEGV (segmentation fault), etc. These are exceptions originating from Glibc/application execution. For example, these 32 sync signals are trap-and-emulate cases for the CPUID instruction, forbidden in SGX enclaves (we know this because we manually inspected this logic). In particular, Glibc initializes itself and checks for different CPU information, and thus invokes CPUID 32 times.
- What are the 0 async signals? These are asynchronous signals forwarded by the host Linux to Gramine. These signals are SIGINT (interrupt), SIGCONT (continue), SIGKILL (the process is going to be killed), SIGCHLD (child process terminated) etc. None of these signals happened during HelloWorld execution.
- Performance counters provide a lot of different information. Here we only show the snippet with number of cycles, instructions, and branches taken. By itself, this particular output is not interesting. In reality, performance counters should be compared against “golden runs” to deduce any interesting trends.
Effects of system calls / ocalls¶
One of the main sources of overhead on modern SGX-enabled Intel processors is a pure software one: enclavized applications must communicate with the outside world, and for this communication they must perform system calls / OCALLs. Every OCALL results in one EEXIT to exit from the enclave to the untrusted world and then one EENTER to re-enter the enclave (unless you are using Exitless, see below). Moreover, OCALLs typically copy some data from the enclave to the outside and vice versa – for example, network and file system OCALLs must copy network packets and files to/from the enclave.
So, there are several sources of overhead that need to be understood with regard to OCALLs:
OCALL = EEXIT + host processing + EENTER. Recall that each EEXIT flushes the CPU caches and possibly invalidates Branch Predictors and TLBs. Similarly, EENTER performs many checks and requires hardware-internal synchronization of cores. Some studies show each EEXIT and EENTER cost around 8,000 – 12,000 cycles (compare it with normal syscalls costing around 100 cycles each). Note that the cost of EENTER/EEXIT depends on the CPU model, its firmware, applied microcode patches, and other platform characteristics.
- OCALLs purge CPU caches. This means that after each OCALL, data that was cached in say L1 data cache is not there anymore. This effectively negates the effect of warm caches in the SGX environment.
- Many OCALLs perform I/O: they copy data to/from the enclave. Copying is obligatory to prevent Time-of-check to time-of-use (TOCTOU) attacks and is dictated by the SGX design. This is an unavoidable tax. In I/O intensive workloads, the overhead of copying may constitute 15-50% over the baseline performance of the native application. For example, databases and web servers copy user requests inside the enclave and copy the results/web pages out. In another example, applications that manipulate files perform a lot of file system I/O, copying data blocks in and out of the enclave.
- OCALLs generally correspond 1:1 to the system calls that the application
performs, but not always. Typical system calls like
send()indeed correspond 1:1 to Gramine’s OCALLs and thus introduce almost no overhead in the code path. However, some system calls are emulated in a more sophisticated way: e.g., Linux-specific
epoll()is emulated via more generic
poll()and this requires some additional logic. Fortunately, such calls are never a real bottleneck in Gramine because they are not on hot paths of applications. Probably the only exceptional system call is
gettimeofday()– and only on older Intel CPUs (see below).
gettimeofday()system call is special. On normal Linux, it is implemented via vDSO and a fast RDTSC instruction. Platforms older than Icelake typically forbid RDTSC inside an SGX enclave (this is a hardware limitation), and so
gettimeofday()falls back to the expensive OCALL. Gramine is smart enough to identify whether the platform supports RDTSC inside enclaves, and uses the fast RDTSC logic to emulate
gettimeofday(). Rule of thumb: if you think that the bottleneck of your deployment is
gettimeofday(), move to a newer (Icelake) processor. If you cannot move to a newer platform, you are limited by SGX hardware (you can try to modify the application itself to issue less gettimeofday’s).
Note this feature is currently insecure and not recommended for production usage (potentially susceptible to CVE-2022-21233 aka INTEL-SA-00657 and CVE-2022-21166 aka INTEL-SA-00615).
Gramine supports the Exitless (or Switchless) feature – it trades off CPU cores for faster OCALL execution. More specifically, with Exitless, enclave threads do not exit the enclave on OCALLs but instead busy wait for untrusted helper threads which perform OCALLs (system calls) on their behalf. Untrusted helper threads are created at Gramine start-up and burn CPU cycles busy waiting for requests for OCALLs from enclave threads (untrusted helper threads periodically sleep if there have been no OCALL requests for a long time to save some CPU cycles).
Exitless is configured by
sgx.insecure__rpc_thread_num = xyz. By default,
the Exitless feature is disabled – all enclave threads perform an actual OCALL
for each system call and exit the enclave. The feature can be disabled by
sgx.insecure__rpc_thread_num = 0.
You must decide how many untrusted helper RPC threads your application needs. A
rule of thumb: specify
sgx.insecure__rpc_thread_num == sgx.max_threads,
i.e., the number of untrusted RPC threads should be the same as the number of
enclave threads. For example, native Redis 6.0 uses 3-4 enclave threads during
its execution, plus Gramine uses another 1-2 helper enclave threads. So Redis
manifest has an over-approximation of this number:
sgx.max_threads = 8. Thus,
to correctly enable the Exitless feature, specify
sgx.insecure__rpc_thread_num = 8. Here is an example:
# exitless disabled: `sgx.max_threads = 8` and `sgx.insecure__rpc_thread_num = 0` CI-Examples/redis$ gramine-sgx redis-server --save '' --protected-mode no & CI-Examples/redis$ src/src/redis-benchmark -t set 43010.75 requests per second # exitless enabled: `sgx.max_threads = 8` and `sgx.insecure__rpc_thread_num = 8` CI-Examples/redis$ gramine-sgx redis-server --save '' --protected-mode no & CI-Examples/redis$ src/src/redis-benchmark -t set 68119.89 requests per second
As you can see, enabling the Exitless feature improves performance of Redis by
58%. This comes at a price: there are now 8 additional threads occupying
additional CPU cores (you can see these additional threads by running
| grep loader while Gramine is running). We recommend to use Exitless only for
single-threaded applications or if you care more about latency than throughput.
We also recommend to use core pinning via taskset or even isolating cores via
isolcpus or disabling interrupts on cores via
nohz_full. It is also
beneficial to put all enclave threads on one set of cores (e.g., on first
hyper-threads if you have hyper-threading enabled on your platform) and all
untrusted RPC threads on another set of cores (e.g., on second hyper-threads).
In general, the classical performance-tuning strategies are applicable for
Gramine and Exitless multi-threaded workloads.
Optional CPU features (AVX, AVX512, MPX, PKRU, AMX)¶
SGX technology allows to specify which CPU features are required to run the SGX
enclave. Gramine “inherits” this and has the following manifest options:
sgx.require_amx. By default, all of them are set to
false – this means that SGX hardware will allow running the SGX enclave on
any system, whether the system has AVX/AVX512/MPX/PKRU/AMX features or not.
Gramine typically correctly identifies the features of the underlying platform
and propagates the information on AVX/AVX512/MPX/PKRU/AMX inside the enclave and
to the application. It is recommended to leave these manifest options as-is (set
false). However, we observed on some platforms that the graminized
application cannot detect these features and falls back to a slow
implementation. For example, some crypto libraries do not recognize AVX on the
platform and use very slow functions, leading to 10-100x overhead over native
(we still don’t know the reason for this behavior). If you suspect this can be
your case, enable the features in the manifest, e.g., set
For more information on SGX logic regarding optional CPU features, see the Intel Software Developer Manual, Table 38-3 (“Layout of ATTRIBUTES Structure”) under the SGX section.
Gramine supports multi-threaded applications. Gramine implements many
optimizations and performance-relevant system calls related to multi-threading
and scheduling policies (e.g.,
Gramine supports multi-process applications, i.e., applications that run as several inter-dependent processes. Typical examples are bash scripts: one main bash script spawns many additional processes to perform some operations. Another typical example is Python: it usually spawns helper processes to obtain system information. Finally, many applications are multi-process by design, e.g., Nginx and Apache web servers spawn multiple worker processes.
For each new child, the parent Gramine process creates a new process with a new Gramine instance and thus a new enclave. For example, if Nginx main process creates 4 workers, then there will be 5 Gramine instances and 5 SGX enclaves: one main Gramine process with its enclave and 4 child Gramine processes with 4 enclaves.
To create a new child process, Linux has the following system calls:
clone(). All these interfaces copy the whole
memory of the parent process into the child, as well as all the resources like
opened files, network connections, etc. In a normal environment, this copying is
very fast because it uses the copy-on-write semantics. However, the SGX hardware
doesn’t have the notions of copy-on-write and sharing of memory. Therefore,
fork/vfork/clone via the checkpoint-and-restore mechanism:
all enclave memory and resources of the parent process are serialized into one
blob of data, the blob is encrypted and sent to the child process. The child
process awaits this blob of data, receives it, decrypts it, and restores into
its own enclave memory. This is a much more expensive operation than
copy-on-write, therefore forking in Gramine is much slower than in native
Linux. Some studies report 1,000x overhead of forking over native.
Moreover, multi-process applications periodically need to communicate with each other. For example, the Nginx parent process sends a signal to one of the worker processes to inform that a new request is available for processing. All this Inter-Process Communication (IPC) is transparently encrypted in Gramine. Encryption by itself incurs 1-10% overhead. This means that a communication-heavy multi-process application may experience significant overheads.
To summarize, there are two sources of overhead for multi-process applications in Gramine:
clone()system calls are very expensive in Gramine and in SGX in general. This is because Intel SGX lacks the mechanisms for memory sharing and copy-on-write semantics. They are emulated via checkpoint-and-restore in Gramine.
- Inter-Process Communication (IPC) is moderately expensive in Gramine because all IPC is transparently encrypted/decrypted using the TLS-PSK with AES-GCM crypto.
Choice of SGX machine¶
Modern Icelake server machines remove many of the hardware bottlenecks of Intel SGX. If you must use an older machine (Skylake, Caby Lake, Mehlow), you should be aware that they have severe SGX-hardware limitations. In particular:
EPC size. You can think of EPC as a physical cache (just like L3 cache) for enclave pages. On older machines (before Icelake servers), EPC is only 128-256MB in size. This means that if the application has a working set size of more than 100-200MB, enclave pages will be evicted from EPC into RAM. Eviction of enclave pages (also called EPC swapping or paging) is a very expensive hardware operation. Some applications have a working set size of MBs/GBs of data, so performance will be significantly impaired.
Note that modern Icelake servers have EPC size of up to 1TB and therefore they are not affected by EPC swapping. A simple way to verify the amount of EPC available on your machine is to execute Gramine’s utility
RDTSC/RDTSCP instructions. These instructions are forbidden to execute in an SGX enclave on older machines. Unfortunately, many applications and runtimes use these instructions frequently, assuming that they are always available. This leads to significant overheads when running such applications: Gramine treats each RDTSC instruction as trap-and-emulate, which is very expensive (enclave performs an AEX, Gramine enters the enclave, fixes RDTSC, exits the enclave, and re-enters it from the interrupted point). Solution: move to newer Intel processors that like Icelake which allow RDTSC inside the enclave.
CPUID and SYSCALL instructions. These instructions are forbidden to execute in an SGX enclave on all currently available machines. Fortunately, applications use these instructions typically only during initialization and never on hot paths. Gramine emulates CPUID and SYSCALL similarly to RDTSC, but since this happens very infrequently, it is not a realistic bottleneck. However, it is always advisable to verify that the application doesn’t rely on CPUID and SYSCALL too much. This is especially important for statically built applications that may rely on raw SYSCALL instructions instead of calling Glibc (recall that Gramine replaces native Glibc with our patched version that performs function calls inside Gramine instead of raw SYSCALL instructions and thus avoids this overhead).
CPU topology. The CPU topology may negatively affect performance of Gramine. For example, if the machine has several NUMA domains, it is important to restrict Gramine runs to only one NUMA domain, e.g., via the command
numactl --cpunodebind=0 --membind=0. Otherwise Gramine may spread enclave threads and enclave memory across several NUMA domains, which will lead to higher memory access latencies and overall worse performance.
Glibc malloc tuning¶
Depending on the number of threads and the value of
might encounter pathological performance due to a combination of various factors.
Specifically, the default settings of glibc’s
malloc assume that virtual memory is
virtually unlimited, and, as an optimization, request a per-thread arena
of 64 MiB from the kernel when a thread first calls
Due to the limitations of SGX v1, we must back each allocation with physical memory
immediately, which breaks the assumption that speculatively allocating 64 MiB is not
a big deal — when many threads are spawned (and call
malloc), the per-thread arenas
might consume a large portion of the memory reserved for the enclave.
When this happens, calls to
malloc won’t fail, as the allocator will
allocate a single page to serve the request instead. However, no attempt will
be made to make use of the rest of the page, wasting most of the memory.
Moreover, glibc will retry allocating the arena each time
called, perhaps in a hope that the memory situation that prevented the previous
attempt from succeeding has since passed.
All together, this means that, unless
64M * (application's thread count) fits
malloc will be much slower and much less
memory-efficient than it should be, on some of the threads involved, because each
call will now cause multiple relatively expensive calls to
mmap, and effectively
round up the request size to a multiple of 4096.
One way to solve this is to limit the number of threads that are allowed to have
their own arena. This can be done with either a call to
mallopt, or an environment
variable set in the manifest:
# Only create one malloc arena. loader.env.MALLOC_ARENA_MAX = "1"
This does have its own performance implications, but the impact is much smaller than the pathological behavior described above.
Another solution would be to set
sgx.enclave_size to a much higher value,
to accomodate each thread creating its own arena. Do keep in mind, however,
that each process spawns its own enclave, so in a multi-process application,
the actual memory consumption will be a multiple of this setting. If the memory
consumption is not a problem for your usecase, you might observe better
performance with this approach than when limiting
For performance testing, always use the non-debug versions of all software. In
particular, build Gramine in non-debug configuration
meson --buildtype=release). Also build the application itself in non-debug
configuration (in the example Makefiles, simple
make SGX=1 defaults to
non-debug). Finally, disable the debug log of Gramine by specifying the manifest
loader.log_level = "none".
There are several manifest options that may improve performance of some workloads. The manifest options include:
libos.check_invalid_pointers = false– disable checks of invalid pointers on system call invocations. Most real-world applications never provide invalid arguments to system calls, so there is no need in additional checks.
sgx.preheat_enclave = true– pre-fault all enclave pages during enclave initialization. This shifts the overhead of page faults on non-present enclave pages from runtime to enclave startup time. Using this option makes sense only if the whole enclave memory fits into EPC and if EDMM is not used (
sgx.edmm_enable = false).
If your application periodically fails and complains about seemingly irrelevant
things, it may be due to insufficient enclave memory. Please try to increase
enclave size by tweaking
sgx.enclave_size = "512M",
sgx.enclave_size = "1G",
sgx.enclave_size = "2G", and so on. If this
doesn’t help, it could be due to insufficient stack size: in this case try to
sys.stack.size = "256K",
sys.stack.size = "2M",
sys.stack.size = "4M" and so on. Finally, if Gramine complains about
insufficient number of TCSs or threads, increase
sgx.max_threads = 4,
sgx.max_threads = 8,
sgx.max_threads = 16, and so on.
Do not forget about the cost of software encryption! Gramine transparently encrypts many means of communication:
- Inter-Process Communication (IPC) is encrypted via TLS-PSK. Regular pipes, FIFO pipes, UNIX domain sockets are all transparently encrypted.
- Files mounted as
type = "encrypted"are transparently encrypted/decrypted on each file access via SGX SDK Merkle-tree format.
Fork/vfork/cloneall require to generate an encrypted checkpoint of the whole enclave memory, send it from parent process to the child, and decrypt it (all via TLS-PSK).
- All SGX attestation, RA-TLS, and Secret Provisioning network communication is encrypted via TLS. Moreover, attestation depends on the internet speed and the remote party, so can also become a bottleneck.
Parsing the manifest can be another source of overhead. If you have a really long manifest (several MBs in size), parsing such a manifest may significantly deteriorate start-up performance. This is rarely a case, but keep manifests as small as possible. Note that this overhead is due to our sub-optimal parser. Once Gramine moves to a better manifest parser, this won’t be an issue.
Finally, recall that by default Gramine doesn’t propagate environment variables
into the SGX enclave. Thus, environment variables like
MKL_NUM_THREADS are not visible to the graminized application by default.
To propagate them into the enclave, either use the insecure manifest option
loader.insecure__use_host_env = true (don’t use this in production!) or
specify them explicitly in the manifest via
loader.env.OMP_NUM_THREADS = "8". Also, it is always better to specify such
environment variables explicitly because a graminized application may determine
the number of available CPUs incorrectly.
This section describes how to use perf, a powerful Linux profiling tool.
perf provided by your distribution¶
perf. It will complain about not having a kernel-specific package, such as
- Install the kernel-specific package.
The above might not work if you have a custom kernel. In that case, you might
want to use the distribution-provided version anyway (install
linux-tools-generic and use
/usr/lib/linux-tools-<VERSION>/perf), but it
might not be fully compatible with your kernel. It might be better to build
You might also want to compile your own
perf to make use of libraries that
the default version is not compiled against.
Building your own
Download the kernel: run
uname -rto check your kernel version, then clone the right branch:
git clone --single-branch --branch linux-5.4.y \ https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
Check the beginning of the output. It will display warnings about missing libraries, and suggest how to install them.
Install the missing ones, depending on the features you need. You will need at least
libunwind-devto get proper symbols and stack trace.
libslang2-devis also nice, as it will enable a terminal UI for some commands.
makeagain and verify that the necessary features have been enabled. (You can also run
ldd perfto check which shared libraries it uses).
Install somewhere, e.g.
sudo make install DESTDIR=/usr/local.
Recording samples with
To record (saves
perf record gramine-direct application
To view the report for
This is useful in non-SGX mode. Unfortunately, in SGX mode, it will not account correctly for the code inside the enclave.
Some useful options for recording (
--call-graph dwarf: collect information about callers
-F 50: collect 50 (or any other number) of samples per second, can be useful to reduce overhead and file size
-e cpu-clock: sample the
cpu-clockevent, which will be triggered also inside enclave (as opposed to the default
cpu-cyclesevent). Unfortunately such events will be counted towards
async_exit_pointerinstead of functions executing inside enclave (but see also SGX profiling).
Some useful options for displaying the report (
--no-children: sort based on “self time”, i.e. time spent in a given function excluding its children (the default is to sort by total time spent in a function).
There is support for profiling the code inside the SGX enclave. Here is how to use it:
Compile Gramine with
You can also use
--buildtype=debugoptimized(optimizations enabled) makes Gramine performance more similar to release build.
Profiling can be done only on debug enclaves. Add
sgx.debug = trueto manifest.
sgx.profile.enable = "main"to manifest (to collect data for the main process), or
sgx.profile.enable = "all"(to collect data for all processes).
sgx.profile.with_stack = truefor call chain information.)
Run your application. It should say something like
Profile data written to sgx-perf.dataon process exit (in case of
sgx.profile.enable = "all", multiple files will be written).
perf report -i <data file>(see Profiling with perf above).
Some applications might run for a long time or forever (e.g. Redis), generating
too much perf data. In such cases, user may want to terminate the application
prematurely. Killing the application abruptly via SIGKILL will result in incorrect
perf data. Instead, add
sys.enable_sigterm_injection = true to manifest and
terminate the application using command
kill <pid> (i.e. send SIGTERM).
Note: The accuracy of this tool is unclear (though we had positive experiences using the tool so far). The SGX profiling works by measuring the value of instruction pointer on each asynchronous enclave exit (AEX), which happen on Linux scheduler interrupts, as well as other events such as page faults. While we attempt to measure time (and not only count occurences), the results might be inaccurate.
It’s also possible to discover what OCALLs are being executed, which should help
attribute the EEXIT numbers given by
sgx.enable_stats. There are two ways to
sgx.profile.mode = "ocall_inner"and
sgx.profile.with_stack = 1. This will give you a report on what enclave code is causing the OCALLs (best viewed with
perf report --no-children).
with_stackoption is important: without it, the report will only show the last function before enclave exit, which is usually the same regardless of which OCALL we’re executing.
sgx.profile.mode = "ocall_outer". This will give you a report on what outer PAL code is handling the OCALLs (
Warning: The report for OCALL modes should be interpreted in term of number
of OCALLs, not time spent in them. The profiler records a sample every time an
OCALL is executed, and
perf report displays percentages based on the number
Profiling SGX hotspots with Intel VTune Profiler¶
This section describes how to use VTune profiler to find SGX hotspots.
Collecting SGX hotspots and viewing the report¶
Compile Gramine with
-Dvtune=enabled -Dvtune_sdk_path=<VTune SDK install path>.
vtune_sdk_pathis not provided, Gramine will use the default VTune installation path.
sgx.vtune_profile = trueand
sgx.debug = trueto the manifest.
Run your application under VTune.
vtune -collect sgx-hotspots -result-dir results -- gramine-sgx <workload>
It will output the data collected to a directory
To view the report, run
vtune -report hotspots -r <vtune data collection output directory>or use Intel VTune Profiler GUI application.